1 systemd System and Service Manager
5 Announcement of Future Feature Removal
7 * Please note that we intend to remove cgroupsv1 support from systemd
8 release after EOY 2023. If you run services that make explicit use of
9 cgroupsv1 features, please implement compatibility with cgroupsv2
10 sooner rather than later, if you haven't done so yet. Most of Linux
11 userspace has been ported over already.
15 Backwards-incompatible changes:
17 * The minimum kernel version required has been bumped from 3.13 to 4.15,
18 and CLOCK_BOOTTIME is now assumed to always exist.
20 * C11 with GNU extensions (aka "gnu11") is now used to build our
21 components. Public API headers are still restricted to ISO C89.
23 * In v250, a systemd-networkd feature that automatically configures
24 routes to addresses specified in AllowedIPs= was added and enabled by
25 default. However, this causes network connectivity issues in many
26 existing setups. Hence, it has been disabled by default since
27 systemd-stable 250.3. The feature can still be used by explicitly
28 configuring RouteTable= setting in .netdev files.
30 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
31 when a Condition*= check does not succeed, restoring the JobRemoved
32 signal to the behaviour it had before v250.
34 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
35 GetImageMetadataWithExtensions() have been fixed to provide an extra
36 return parameter, containing the actual extension release metadata.
37 The current implementation was judged to be broken and unusable, and
38 thus the usual procedure of adding a new set of methods was skipped,
39 and backward compatibility broken instead on the assumption that
40 nobody can be affected given the current state of this interface.
42 * All kernels supported by systemd mix bytes returned by RDRAND (or
43 similar) into the entropy pool at early boot. This means that on
44 those systems, even if /dev/urandom is not yet initialized, it still
45 returns bytes that are of at least RDRAND quality. For that reason,
46 we no longer have reason to invoke RDRAND from systemd itself, which
47 has historically been a source of bugs. Furthermore, kernels ≥5.6
48 provide the getrandom(GRND_INSECURE) interface for returning random
49 bytes before the entropy pool is initialized without warning into
50 kmsg, which is what we attempt to use if available. systemd's direct
51 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
52 running an older kernel may experience kmsg warnings that were not
53 seen with 250. For newer kernels, non-x86 systems, or older x86
54 systems, there should be no visible changes.
56 * sd-boot will now measure the kernel command line into TPM PCR 12
57 rather than PCR 8. This improves usefulness of the measurements on
58 systems where sd-boot is chainloaded from Grub. Grub measures all
59 commands its executes into PCR 8, which makes it very hard to use
60 reasonably, hence separate ourselves from that and use PCR 12
61 instead, which is what certain Ubuntu editions already do. To retain
62 compatibility with systems running older systemd systems a new meson
63 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
64 If enabled, the measurement is done twice: into the new-style PCR 12
65 *and* the old-style PCR 8. It's strongly advised to migrate all users
66 to PCR 12 for this purpose in the long run, as we intend to remove
67 this compatibility feature in two years' time.
69 * busctl capture now writes output in the newer pcapng format instead
72 * A udev rule that imported hwdb matches for USB devices with lowercase
73 hexadecimal vendor/product ID digits was added in systemd 250. This
74 has been reverted, since uppercase hexadecimal digits are supposed to
75 be used, and we already had a rule with the appropriate match.
77 Users might need to adjust their local hwdb entries.
79 * arch_prctl(2) has been moved to the @default set in the syscall filters
80 (as exposed via the SystemCallFilter= setting in service unit files).
81 It is apparently used by the linker now.
83 * The tmpfiles entries that create the /run/systemd/netif directory and
84 its subdirectories were moved from tmpfiles.d/systemd.conf to
85 tmpfiles.d/systemd-network.conf.
87 Users might need to adjust their files that override tmpfiles.d/systemd.conf
88 to account for this change.
90 * The requirement for Portable Services images to contain a well-formed
91 os-release file (i.e.: contain at least an ID field) is now enforced.
92 This applies to base images and extensions, and also to systemd-sysext.
94 Changes in the Boot Loader Specification, kernel-install and sd-boot:
96 * kernel-install's and bootctl's Boot Loader Specification Type #1
97 entry generation logic has been reworked. The user may now pick
98 explicitly by which "token" string to name the installation's boot
99 entries, via the new /etc/kernel/entry-token file or the new
100 --entry-token= switch to bootctl. By default — as before — the
101 entries are named after the local machine ID. However, in "golden
102 image" environments, where the machine ID shall be initialized on
103 first boot (as opposed to at installation time before first boot) the
104 machine ID will not be available at build time. In this case the
105 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
106 file) may be used to override the "token" for the entries, for
107 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
108 make the OS images independent of any machine ID, and ensure that the
109 images will not carry any identifiable information before first boot,
110 but on the other hand means that multiple parallel installations of
111 the very same image on the same disk cannot be supported.
113 Summary: if you are building golden images that shall acquire
114 identity information exclusively on first boot, make sure to both
115 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
116 value of the IMAGE_ID= or ID= field of /etc/os-release or another
117 suitable identifier before deploying the image.
119 * The Boot Loader Specification has been extended with
120 /loader/entries.srel file located in the EFI System Partition (ESP)
121 that disambiguates the format of the entries in the /loader/entries/
122 directory (in order to discern them from incompatible uses of this
123 directory by other projects). For entries that follow the
124 Specification, the string "type1" is stored in this file.
126 bootctl will now write this file automatically when installing the
127 systemd-boot boot loader.
129 * kernel-install supports a new initrd_generator= setting in
130 /etc/kernel/install.conf, that is exported as
131 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
132 allows choosing different initrd generators.
134 * kernel-install will now create a "staging area" (an initially-empty
135 directory to gather files for a Boot Loader Specification Type #1
136 entry). The path to this directory is exported as
137 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
138 drop files there instead of writing them directly to the final
139 location. kernel-install will move them when all files have been
140 prepared successfully.
142 * New option sort-key= has been added to the Boot Loader Specification
143 to override the sorting order of the entries in the boot menu. It is
144 read by sd-boot and bootctl, and will be written by kernel-install,
145 with the default value of IMAGE_ID= or ID= fields from
146 os-release. Together, this means that on multiboot installations,
147 entries should be grouped and sorted in a predictable way.
149 * The sort order of boot entries has been updated: entries which have
150 the new field sort-key= are sorted by it first, and all entries
151 without it are ordered later. After that, entries are sorted by
152 version so that newest entries are towards the beginning of the list.
154 * The kernel-install tool gained a new 'inspect' verb which shows the
155 paths and other settings used.
157 * sd-boot can now optionally beep when the menu is shown and menu
158 entries are selected, which can be useful on machines without a
159 working display. (Controllable via a loader.conf setting.)
161 * The --make-machine-id-directory= switch to bootctl has been replaced
162 by --make-entry-directory=, given that the entry directory is not
163 necessarily named after the machine ID, but after some other suitable
164 ID as selected via --entry-token= described above. The old name of
165 the option is still understood to maximize compatibility.
167 * 'bootctl list' gained support for a new --json= switch to output boot
168 menu entries in JSON format.
170 * 'bootctl is-installed' now supports the --graceful, and various verbs
171 omit output with the new option --quiet.
173 Changes in systemd-homed:
175 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
176 of activated home directories it manages (if the kernel and selected
177 file systems support it). So far it mapped three UID ranges: the
178 range from 0…60000, the user's own UID, and the range 60514…65534,
179 leaving everything else unmapped (in other words, the 16bit UID range
180 is mapped almost fully, with the exception of the UID subrange used
181 for systemd-homed users, with one exception: the user's own UID).
182 Unmapped UIDs may not be used for file ownership in the home
183 directory — any chown() attempts with them will fail. With this
184 release a fourth range is added to these mappings:
185 524288…1879048191. This range is the UID range intended for container
188 https://systemd.io/UIDS-GIDS
190 This range may be used for container managers that place container OS
191 trees in the home directory (which is a questionable approach, for
192 quota, permission, SUID handling and network file system
193 compatibility reasons, but nonetheless apparently commonplace). Note
194 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
195 UID assignments from the range are not managed or mapped by
196 `systemd-homed`, and must be managed with other mechanisms, in the
197 context of the local system.
199 Typically, a better approach to user namespacing in relevant
200 container managers would be to leave container OS trees on disk at
201 UID offset 0, but then map them to a dynamically allocated runtime
202 UID range via another UID mount map at container invocation
203 time. That way user namespace UID ranges become strictly a runtime
204 concept, and do not leak into persistent file systems, persistent
205 user databases or persistent configuration, thus greatly simplifying
206 handling, and improving compatibility with home directories intended
207 to be portable like the ones managed by systemd-homed.
209 Changes in shared libraries:
211 * A new libsystemd-core-<version>.so private shared library is
212 installed under /usr/lib/systemd/system, mirroring the existing
213 libsystemd-shared-<version>.so library. This allows the total
214 installation size to be reduced by binary code reuse.
216 * The <version> tag used in the name of libsystemd-shared.so and
217 libsystemd-core.so can be configured via the meson option
218 'shared-lib-tag'. Distributions may build subsequent versions of the
219 systemd package with unique tags (e.g. the full package version),
220 thus allowing multiple installations of those shared libraries to be
221 available at the same time. This is intended to fix an issue where
222 programs that link to those libraries would fail to execute because
223 they were installed earlier or later than the appropriate version of
226 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
227 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
228 format instead of as a simple series of hex characters.
230 * The sd-device API gained two new calls sd_device_new_from_devname()
231 and sd_device_new_from_path() which permit allocating an sd_device
232 object from a device node name or file system path.
234 * sd-device also gained a new call sd_device_open() which will open the
235 device node associated with a device for which an sd_device object
236 has been allocated. The call is supposed to address races around
237 device nodes being removed/recycled due to hotplug events, or media
238 change events: the call checks internally whether the major/minor of
239 the device node and the "diskseq" (in case of block devices) match
240 with the metadata loaded in the sd_device object, thus ensuring that
241 the device once opened really matches the provided sd_device object.
243 Changes in PID1, systemctl, and systemd-oomd:
245 * A new set of service monitor environment variables will be passed to
246 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
247 handler unit as OnFailure=/OnSuccess=. The variables are:
248 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
249 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
250 handler needs to watch multiple units, use a templated handler.
252 * A new ExtensionDirectories= setting in service unit files allows
253 system extensions to be loaded from a directory. (It is similar to
254 ExtensionImages=, but takes paths to directories, instead of
257 'portablectl attach --extension=' now also accepts directory paths.
259 * The user.delegate and user.invocation_id extended attributes on
260 cgroups are used in addition to trusted.delegate and
261 trusted.invocation_id. The latter pair requires privileges to set,
262 but the former doesn't and can be also set by the unprivileged user
265 (Only supported on kernels ≥5.6.)
267 * Units that were killed by systemd-oomd will now have a service result
268 of 'oom-kill'. The number of times a service was killed is tallied
269 in the 'user.oomd_ooms' extended attribute.
271 The OOMPolicy= unit file setting is now also honoured by
274 * In unit files the new %y/%Y specifiers can be used to refer to
275 normalized unit file path, which is particularly useful for symlinked
278 The new %q specifier resolves to the pretty hostname
279 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
281 The new %d specifier resolves to the credentials directory of a
282 service (same as $CREDENTIALS_DIRECTORY).
284 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
285 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
286 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
287 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
288 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
289 MountFlags= service settings now also work in unprivileged user
290 services, i.e. those run by the user's --user service manager, as long
291 as user namespaces are enabled on the system.
293 * Services with Restart=always and a failing ExecCondition= will no
294 longer be restarted, to bring ExecCondition= behaviour in line with
295 Condition*= settings.
297 * LoadCredential= now accepts a directory as the argument; all files
298 from the directory will be loaded as credentials.
300 * A new D-Bus property ControlGroupId is now exposed on service units,
301 that encapsulates the service's numeric cgroup ID that newer kernels
302 assign to each cgroup.
304 * PID 1 gained support for configuring the "pre-timeout" of watchdog
305 devices and the associated governor, via the new
306 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
307 options in /etc/systemd/system.conf.
309 * systemctl's --timestamp= option gained a new choice "unix", to show
310 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
312 * A new "taint" flag named "old-kernel" is introduced which is set when
313 the kernel systemd runs on is older then the current baseline version
314 (see above). The flag is shown in "systemctl status" output.
316 * Two additional taint flags "short-uid-range" and "short-gid-range"
317 have been added as well, which are set when systemd notices it is run
318 within a userns namespace that does not define the full 0…65535 UID
321 * A new "unmerged-usr" taint flag has been added that is set whenever
322 running on systems where /bin/ + /sbin/ are *not* symlinks to their
323 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
326 * Generators invoked by PID 1 will now have a couple of useful
327 environment variables set describing the execution context a
328 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
329 system service manager, or from the per-user service
330 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
331 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
332 systemd considers the current boot to be a "first"
333 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
334 detected and which type of hypervisor/container
335 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
338 * PID 1 will now automatically pick up system credentials from qemu's
339 fw_cfg interface, thus allowing passing arbitrary data into VM
340 systems similar to how this is already supported for passing them
341 into `systemd-nspawn` containers. Credentials may now also be passed
342 in via the new kernel command line option `systemd.set_credential=`
343 (note that kernel command line options are world-readable during
344 runtime, and only useful for credentials that require no
345 confidentiality). The credentials that can be passed to unified
346 kernels that use the `systemd-stub` UEFI stub are now similarly
347 picked up automatically. Automatic importing of system credentials
348 this way can be turned off via the new
349 `systemd.import_credentials=no` kernel command line option.
351 * LoadCredential= will now automatically look for credentials in the
352 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
353 the argument is not an absolute path. Similarly,
354 LoadCredentialEncrypted= will check the same directories plus
355 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
356 /usr/lib/credstore.encrypted/. The idea is to use those directories
357 as the system-wide location for credentials that services should pick
360 * System and service credentials are described in great detail in a new
363 https://systemd.io/CREDENTIALS
365 Changes in systemd-journald:
367 * The journal JSON export format has been added to listed of stable
368 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
370 * journalctl --list-boots now supports JSON output and the --reverse option.
372 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
373 updated, BUILDING_IMAGES is new:
375 https://systemd.io/JOURNAL_EXPORT_FORMATS
376 https://systemd.io/BUILDING_IMAGES
380 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
381 calculators, etc.), the other AV production devices (DJ tables,
382 keypads, etc.) that should accessible to the seat owner user by
385 * udevadm trigger gained a new --prioritized-subsystem= option to
386 process certain subsystems (and all their parent devices) earlier.
388 systemd-udev-trigger.service now uses this new option to trigger
389 block and TPM devices first, hopefully making the boot a bit faster.
391 * udevadm trigger now implements --type=all, --initialized-match,
392 --initialized-nomatch to trigger both subsystems and devices, only
393 already-initialized devices, and only devices which haven't been
394 initialized yet, respectively.
396 * udevadm gained a new "wait" command for safely waiting for a specific
397 device to show up in the udev device database. This is useful in
398 scripts that asynchronously allocate a block device (e.g. through
399 repartitioning, or allocating a loopback device or similar) and need
400 to synchronize on the creation to complete.
402 * udevadm gained a new "lock" command for locking one or more block
403 devices while formatting it or writing a partition table to it. It is
404 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
405 usable in scripts dealing with block devices.
407 * udevadm info will show a couple of additional device fields in its
408 output, and will not apply a limited set of coloring to line types.
410 * udevadm info --tree will now show a tree of objects (i.e. devices and
411 suchlike) in the /sys/ hierarchy.
413 * Block devices will now get a new set of device symlinks in
414 /dev/disk/by-diskseq/<nr>, which may be used to reference block
415 device nodes via the kernel's "diskseq" value. Note that this does
416 not guarantee that opening a device by a symlink like this will
417 guarantee that the opened device actually matches the specified
418 diskseq value. To be safe against races, the actual diskseq value of
419 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
420 the one in the symlink path.
422 * .link files gained support for setting MDI/MID-X on a link.
424 * .link files gained support for [Match] Firmware= setting to match on
425 the device firmware description string. By mistake, it was previously
426 only supported in .network files.
428 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
429 and [SR-IOV] section to configure SR-IOV virtual functions.
431 Changes in systemd-networkd:
433 * The default scope for unicast routes configured through [Route]
434 section is changed to "link", to make the behavior consistent with
435 "ip route" command. The manual configuration of [Route] Scope= is
438 * A new unit systemd-networkd-wait-online@<interface>.service has been
439 added that can be used to wait for a specific network interface to be
442 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
443 that configures the eponymous kernel attribute on the bridge.
445 * .netdev files now can be used to create virtual WLAN devices, and
446 configure various settings on them, via the [WLAN] section.
448 * .link/.network files gained support for [Match] Kind= setting to match
449 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
451 This value is also shown by 'networkctl status'.
453 * The Local= setting in .netdev files for various virtual network
454 devices gained support for specifying, in addition to the network
455 address, the name of a local interface which must have the specified
458 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
459 files, to configure tunnels in external mode (a.k.a. collect metadata
462 * [Network] L2TP= setting was removed. Please use interface specifier in
463 Local= setting in .netdev files of corresponding L2TP interface.
465 * New [DHCPServer] BootServerName=, BootServerAddress=, and
466 BootFilename= settings can be used to configure the server address,
467 server name, and file name sent in the DHCP packet (e.g. to configure
470 Changes in systemd-resolved:
472 * systemd-resolved is started earlier (in sysinit.target), so it
473 available earlier and will also be started in the initrd if installed
476 Changes in disk encryption:
478 * systemd-cryptenroll can now control whether to require the user to
479 enter a PIN when using TPM-based unlocking of a volume via the new
480 --tpm2-with-pin= option.
482 Option tpm2-pin= can be used in /etc/crypttab.
484 * When unlocking devices via TPM, TPM2 parameter encryption is now
485 used, to ensure that communication between CPU and discrete TPM chips
486 cannot be eavesdropped to acquire disk encryption keys.
488 * A new switch --fido2-credential-algorithm= has been added to
489 systemd-cryptenroll allowing selection of the credential algorithm to
490 use when binding encryption to FIDO2 tokens.
492 Changes in systemd-hostnamed:
494 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
495 to override the values gleaned from the hwdb.
497 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
498 /sys/class/dmi/id) to override the chassis that is reported by
501 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
502 for reading the hardware serial number, as reportd by DMI. It also
503 exposes a new method D-Bus property FirmwareVersion that encode the
504 firmware version of the system.
506 Changes in other components:
508 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
509 handling with the values that were configured during systemd build
510 (if /etc/locale.conf has not been created through some other
511 mechanism). This means that /etc/locale.conf should always have
512 reasonable contents and we avoid a potential mismatch in defaults.
514 * The userdbctl tool will now show UID range information as part of the
517 * A new build-time configuration setting default-user-shell= can be
518 used to set the default shell for user records and nspawn shell
519 invocations (instead of of the default /bin/bash).
521 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
522 information dynamically at runtime via IPC.
524 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
525 whether a functioning TPM2 infrastructure is available, i.e. if
526 firmware, kernel driver and systemd all have TPM2 support enabled and
529 * The systemd-creds tool gained support for generating encrypted
530 credentials that are using an empty encryption key. While this
531 provides no integrity nor confidentiality it's useful to implement
532 codeflows that work the same on TPM-ful and TPM2-less systems. The
533 service manager will only accept credentials "encrypted" that way if
534 a TPM2 device cannot be detected, to ensure that credentials
535 "encrypted" like that cannot be used to trick TPM2 systems.
537 * When deciding whether to colorize output, all systemd programs now
538 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
541 * Meson's new install_tag feature is now in use for several components,
542 allowing to build and install select binaries only: pam, nss, devel
543 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
544 $ meson build systemd-boot
545 $ meson install --tags systemd-boot --no-rebuild
546 https://mesonbuild.com/Installing.html#installation-tags
548 * A new build configuration option has been added, to allow selecting the
549 default compression algorithm used by systemd-journald and systemd-coredump.
550 This allows to build-in support for decompressing all supported formats,
551 but choose a specific one for compression. E.g.:
552 $ meson -Ddefault-compression=xz
554 Experimental features:
556 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
557 loader.conf that implements booting Microsoft Windows from the
558 sd-boot in a way that first reboots the system, to reset the TPM
559 PCRs. This improves compatibility with BitLocker's TPM use, as the
560 PCRs will only record the Windows boot process, and not sd-boot
561 itself, thus retaining the PCR measurements not involving sd-boot.
562 Note that this feature is experimental for now, and is likely going
563 to be generalized and renamed in a future release, without retaining
564 compatibility with the current implementation.
566 * A new systemd-sysupdate component has been added that automatically
567 discovers, downloads, and installs A/B-style updates for the host
568 installation itself, or container images, portable service images,
569 and other assets. See the new systemd-sysupdate man page for updates.
571 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
572 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
573 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
574 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
575 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
576 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
577 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
578 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
579 David, David Bond, Davide Cavalca, David Tardon, davijosw,
580 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
581 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
582 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
583 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
584 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
585 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
586 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
587 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
588 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
589 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
590 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
591 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
592 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
593 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
594 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
595 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
596 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
597 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
598 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
599 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
600 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
601 Simon Ellmann, Sonali Srivastava, Stefan Seering,
602 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
603 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
604 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
605 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
606 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
607 Zbigniew Jędrzejewski-Szmek, наб
609 — Edinburgh, 2022-05-21
613 * Support for encrypted and authenticated credentials has been added.
614 This extends the credential logic introduced with v247 to support
615 non-interactive symmetric encryption and authentication, based on a
616 key that is stored on the /var/ file system or in the TPM2 chip (if
617 available), or the combination of both (by default if a TPM2 chip
618 exists the combination is used, otherwise the /var/ key only). The
619 credentials are automatically decrypted at the moment a service is
620 started, and are made accessible to the service itself in unencrypted
621 form. A new tool 'systemd-creds' encrypts credentials for this
622 purpose, and two new service file settings LoadCredentialEncrypted=
623 and SetCredentialEncrypted= configure such credentials.
625 This feature is useful to store sensitive material such as SSL
626 certificates, passwords and similar securely at rest and only decrypt
627 them when needed, and in a way that is tied to the local OS
628 installation or hardware.
630 * systemd-gpt-auto-generator can now automatically set up discoverable
631 LUKS2 encrypted swap partitions.
633 * The GPT Discoverable Partitions Specification has been substantially
634 extended with support for root and /usr/ partitions for the majority
635 of architectures systemd supports. This includes platforms that do
636 not natively support UEFI, because even though GPT is specified under
637 UEFI umbrella, it is useful on other systems too. Specifically,
638 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
639 Portable Services use the concept without requiring UEFI.
641 * The GPT Discoverable Partitions Specifications has been extended with
642 a new set of partitions that may carry PKCS#7 signatures for Verity
643 partitions, encoded in a simple JSON format. This implements a simple
644 mechanism for building disk images that are fully authenticated and
645 can be tested against a set of cryptographic certificates. This is
646 now implemented for the various systemd tools that can operate with
647 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
648 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
649 The PKCS#7 signatures are passed to the kernel (where they are
650 checked against certificates from the kernel keyring), or can be
651 verified against certificates provided in userspace (via a simple
652 drop-in file mechanism).
654 * systemd-dissect's inspection logic will now report for which uses a
655 disk image is intended. Specifically, it will display whether an
656 image is suitable for booting on UEFI or in a container (using
657 systemd-nspawn's --image= switch), whether it can be used as portable
658 service, or attached as system extension.
660 * The system-extension.d/ drop-in files now support a new field
661 SYSEXT_SCOPE= that may encode which purpose a system extension image
662 is for: one of "initrd", "system" or "portable". This is useful to
663 make images more self-descriptive, and to ensure system extensions
664 cannot be attached in the wrong contexts.
666 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
667 be used in portable service images to indicate which unit prefixes
670 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
671 now is able to decode images for non-native architectures as well.
672 This allows systemd-nspawn to boot images of non-native architectures
673 if the corresponding user mode emulator is installed and
674 systemd-binfmtd is running.
676 * systemd-logind gained new settings HandlePowerKeyLongPress=,
677 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
678 HandleHibernateKeyLongPress= which may be used to configure actions
679 when the relevant keys are pressed for more than 5s. This is useful
680 on devices that only have hardware for a subset of these keys. By
681 default, if the reboot key is pressed long the poweroff operation is
682 now triggered, and when the suspend key is pressed long the hibernate
683 operation is triggered. Long pressing the other two keys currently
684 does not trigger any operation by default.
686 * When showing unit status updates on the console during boot and
687 shutdown, and a service is slow to start so that the cylon animation
688 is shown, the most recent sd_notify() STATUS= text is now shown as
689 well. Services may use this to make the boot/shutdown output easier
690 to understand, and to indicate what precisely a service that is slow
691 to start or stop is waiting for. In particular, the per-user service
692 manager instance now reports what it is doing and which service it is
693 waiting for this way to the system service manager.
695 * The service manager will now re-execute on reception of the
696 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
697 only when running as PID 1. There was no signal to request this when
698 running as per-user service manager, i.e. as any other PID than 1.
699 SIGRTMIN+25 works for both system and user managers.
701 * The hardware watchdog logic in PID 1 gained support for operating
702 with the default timeout configured in the hardware, instead of
703 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
704 request this behavior.
706 * A new kernel command line option systemd.watchdog_sec= is now
707 understood which may be used to override the hardware watchdog
708 time-out for the boot.
710 * A new setting DefaultOOMScoreAdjust= is now supported in
711 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
712 to set the default process OOM score adjustment value for processes
713 started by the service manager. For per-user service managers this
714 now defaults to 100, but for per-system service managers is left as
715 is. This means that by default now services forked off the user
716 service manager are more likely to be killed by the OOM killer than
717 system services or the managers themselves.
719 * A new per-service setting RestrictFileSystems= as been added that
720 restricts the file systems a service has access to by their type.
721 This is based on the new BPF LSM of the Linux kernel. It provides an
722 effective way to make certain API file systems unavailable to
723 services (and thus minimizing attack surface). A new command
724 "systemd-analyze filesystems" has been added that lists all known
725 file system types (and how they are grouped together under useful
728 * Services now support a new setting RestrictNetworkInterfaces= for
729 restricting access to specific network interfaces.
731 * Service unit files gained new settings StartupAllowedCPUs= and
732 StartupAllowedMemoryNodes=. These are similar to their counterparts
733 without the "Startup" prefix and apply during the boot process
734 only. This is useful to improve boot-time behavior of the system and
735 assign resources differently during boot than during regular
736 runtime. This is similar to the preexisting StartupCPUWeight=
739 * Related to this: the various StartupXYZ= settings
740 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
741 during shutdown. The settings not prefixed with "Startup" hence apply
742 during regular runtime, and those that are prefixed like that apply
743 during boot and shutdown.
745 * A new per-unit set of conditions/asserts
746 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
747 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
748 pressure is above the configured threshold, using the kernel PSI
749 feature. For more details see systemd.unit(5) and
750 https://docs.kernel.org/accounting/psi.html
752 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
753 ProtectKernelLogs=yes can now be used.
755 * The default maximum numbers of inodes have been raised from 64k to 1M
756 for /dev/, and from 400k to 1M for /tmp/.
758 * The per-user service manager learnt support for communicating with
759 systemd-oomd to acquire OOM kill information.
761 * A new service setting ExecSearchPath= has been added that allows
762 changing the search path for executables for services. It affects
763 where we look for the binaries specified in ExecStart= and similar,
764 and the specified directories are also added the $PATH environment
765 variable passed to invoked processes.
767 * A new setting RuntimeRandomizedExtraSec= has been added for service
768 and scope units that allows extending the runtime time-out as
769 configured by RuntimeMaxSec= with a randomized amount.
771 * The syntax of the service unit settings RuntimeDirectory=,
772 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
773 if the specified value is now suffixed with a colon, followed by
774 another filename, the latter will be created as symbolic link to the
775 specified directory. This allows creating these service directories
776 together with alias symlinks to make them available under multiple
779 * Service unit files gained two new settings TTYRows=/TTYColumns= for
780 configuring rows/columns of the TTY device passed to
781 stdin/stdout/stderr of the service. This is useful to propagate TTY
782 dimensions to a virtual machine.
784 * A new service unit file setting ExitType= has been added that
785 specifies when to assume a service has exited. By default systemd
786 only watches the main process of a service. By setting
787 ExitType=cgroup it can be told to wait for the last process in a
790 * Automount unit files gained a new setting ExtraOptions= that can be
791 used to configure additional mount options to pass to the kernel when
792 mounting the autofs instance.
794 * "Urlification" (generation of ESC sequences that generate clickable
795 hyperlinks in modern terminals) may now be turned off altogether
798 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
799 settings that default to 200 and 2 s respectively. The ratelimit
800 ensures that a path unit cannot cause PID1 to busy-loop when it is
801 trying to trigger a service that is skipped because of a Condition*=
802 not being satisfied. This matches the configuration and behaviour of
805 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
806 as a plug-in for cryptsetup. This means the plain cryptsetup command
807 may now be used to unlock volumes set up this way.
809 * The TPM2 logic in cryptsetup will now automatically detect systems
810 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
811 updates the SHA1 banks. In such a case PCR policies will be
812 automatically bound to the latter, not the former. This makes the PCR
813 policies reliable, but of course do not provide the same level of
814 trust as SHA256 banks.
816 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
817 RSA primary keys in addition to ECC, improving compatibility with
818 TPM2 chips that do not support ECC. RSA keys are much slower to use
819 than ECC, and hence are only used if ECC is not available.
821 * /etc/crypttab gained support for a new token-timeout= setting for
822 encrypted volumes that allows configuration of the maximum time to
823 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
824 the logic will query the user for a regular passphrase/recovery key
827 * Support for activating dm-integrity volumes at boot via a new file
828 /etc/integritytab and the tool systemd-integritysetup have been
829 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
830 with dm-integrity instead of dm-crypt/dm-verity.
832 * The systemd-veritysetup-generator now understands a new usrhash=
833 kernel command line option for specifying the Verity root hash for
834 the partition backing the /usr/ file system. A matching set of
835 systemd.verity_usr_* kernel command line options has been added as
836 well. These all work similar to the corresponding options for the
839 * The sd-device API gained a new API call sd_device_get_diskseq() to
840 return the DISKSEQ property of a device structure. The "disk
841 sequence" concept is a new feature recently introduced to the Linux
842 kernel that allows detecting reuse cycles of block devices, i.e. can
843 be used to recognize when loopback block devices are reused for a
844 different purpose or CD-ROM drives get their media changed.
846 * A new unit systemd-boot-update.service has been added. If enabled
847 (the default) and the sd-boot loader is detected to be installed, it
848 is automatically updated to the newest version when out of date. This
849 is useful to ensure the boot loader remains up-to-date, and updates
850 automatically propagate from the OS tree in /usr/.
852 * sd-boot will now build with SBAT by default in order to facilitate
853 working with recent versions of Shim that require it to be present.
855 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
856 This is used to robustly generate boot entry titles for Windows.
858 * A new generic target unit factory-reset.target has been added. It is
859 hooked into systemd-logind similar in fashion to
860 reboot/poweroff/suspend/hibernate, and is supposed to be used to
861 initiate a factory reset operation. What precisely this operation
862 entails is up for the implementer to decide, the primary goal of the
863 new unit is provide a framework where to plug in the implementation
864 and how to trigger it.
866 * A new meson build-time option 'clock-valid-range-usec-max' has been
867 added which takes a time in µs and defaults to 15 years. If the RTC
868 time is noticed to be more than the specified time ahead of the
869 built-in epoch of systemd (which by default is the release timestamp
870 of systemd) it is assumed that the RTC is not working correctly, and
871 the RTC is reset to the epoch. (It already is reset to the epoch when
872 noticed to be before it.) This should increase the chance that time
873 doesn't accidentally jump too far ahead due to faulty hardware or
876 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
877 which may be used to automatically save the current system time to
878 disk in regular intervals. This is useful to maintain a roughly
879 monotonic clock even without RTC hardware and with some robustness
880 against abnormal system shutdown.
882 * systemd-analyze verify gained support for a pair of new --image= +
883 --root= switches for verifying units below a specific root
884 directory/image instead of on the host.
886 * systemd-analyze verify gained support for verifying unit files under
887 an explicitly specified unit name, independently of what the filename
890 * systemd-analyze verify gained a new switch --recursive-errors= which
891 controls whether to only fail on errors found in the specified units
892 or recursively any dependent units.
894 * systemd-analyze security now supports a new --offline mode for
895 analyzing unit files stored on disk instead of loaded units. It may
896 be combined with --root=/--image to analyze unit files under a root
897 directory or disk image. It also learnt a new --threshold= parameter
898 for specifying an exposure level threshold: if the exposure level
899 exceeds the specified value the call will fail. It also gained a new
900 --security-policy= switch for configuring security policies to
901 enforce on the units. A policy is a JSON file that lists which tests
902 shall be weighted how much to determine the overall exposure
903 level. Altogether these new features are useful for fully automatic
904 analysis and enforcement of security policies on unit files.
906 * systemd-analyze security gain a new --json= switch for JSON output.
908 * systemd-analyze learnt a new --quiet switch for reducing
909 non-essential output. It's honored by the "dot", "syscall-filter",
910 "filesystems" commands.
912 * systemd-analyze security gained a --profile= option that can be used
913 to take into account a portable profile when analyzing portable
914 services, since a lot of the security-related settings are enabled
917 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
918 files, binaries and executables and prints metadata information,
919 including the build-id and other info described on:
920 https://systemd.io/COREDUMP_PACKAGE_METADATA/
922 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
923 section, for automatically propagating DNS settings from other
926 * The static lease DHCP server logic in systemd-networkd may now serve
927 IP addresses outside of the configured IP pool range for the server.
929 * CAN support in systemd-networkd gained four new settings Loopback=,
930 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
931 control modes. It gained a number of further settings for tweaking
934 * The [CAN] section in .network file gained new TimeQuantaNSec=,
935 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
936 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
937 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
938 DataSyncJumpWidth= settings to control bit-timing processed by the
941 * DHCPv4 client support in systemd-networkd learnt a new Label= option
942 for configuring the address label to apply to configure IPv4
945 * The [IPv6AcceptRA] section of .network files gained support for a new
946 UseMTU= setting that may be used to control whether to apply the
947 announced MTU settings to the local interface.
949 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
950 setting to control whether the DHCPv4 client request and process the
953 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
954 [DHCPPrefixDelegation], as now the prefix delegation is also supported
955 with DHCPv4 protocol by enabling the Use6RD= setting.
957 * The [DHCPPrefixDelegation] section in .network file gained a new
958 setting UplinkInterface= to specify the upstream interface.
960 * The [DHCPv6] section in .network file gained a new setting
961 UseDelegatedPrefix= to control whether the delegated prefixes will be
962 propagated to the downstream interfaces.
964 * The [IPv6AcceptRA] section of .network files now understands two new
965 settings UseGateway=/UseRoutePrefix= for explicitly configuring
966 whether to use the relevant fields from the IPv6 Router Advertisement
969 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
970 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
971 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
972 [IPv6AcceptRA] section to control when the DHCPv6 client is started
973 and how the delegated prefixes are handled by the DHCPv6 client.
975 * The IPv6Token= section in the [Network] section is deprecated, and
976 the [IPv6AcceptRA] section gained the Token= setting for its
977 replacement. The [IPv6Prefix] section also gained the Token= setting.
978 The Token= setting gained 'eui64' mode to explicitly configure an
979 address with the EUI64 algorithm based on the interface MAC address.
980 The 'prefixstable' mode can now optionally take a secret key. The
981 Token= setting in the [DHCPPrefixDelegation] section now supports all
982 algorithms supported by the same settings in the other sections.
984 * The [RoutingPolicyRule] section of .network file gained a new
985 SuppressInterfaceGroup= setting.
987 * The IgnoreCarrierLoss= setting in the [Network] section of .network
988 files now allows a duration to be specified, controlling how long to
989 wait before reacting to carrier loss.
991 * The [DHCPServer] section of .network file gained a new Router=
992 setting to specify the router address.
994 * The [CAKE] section of .network files gained various new settings
995 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
996 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
997 and UseRawPacketSize= for configuring CAKE.
999 * systemd-networkd now ships with new default .network files:
1000 80-container-vb.network which matches host-side network bridge device
1001 created by systemd-nspawn's --network-bridge or --network-zone
1002 switch, and 80-6rd-tunnel.network which matches automatically created
1003 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
1005 * systemd-networkd's handling of Endpoint= resolution for WireGuard
1006 interfaces has been improved.
1008 * systemd-networkd will now automatically configure routes to addresses
1009 specified in AllowedIPs=. This feature can be controlled via
1010 RouteTable= and RouteMetric= settings in [WireGuard] or
1011 [WireGuardPeer] sections.
1013 * systemd-networkd will now once again automatically generate persistent
1014 MAC addresses for batadv and bridge interfaces. Users can disable this
1015 by using MACAddress=none in .netdev files.
1017 * systemd-networkd and systemd-udevd now support IP over InfiniBand
1018 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
1019 systemd.netdev files gained the [IPoIB] section.
1021 * systemd-networkd and systemd-udevd now support net.ifname-policy=
1022 option on the kernel command-line. This is implemented through the
1023 systemd-network-generator service that automatically generates
1024 appropriate .link, .network, and .netdev files.
1026 * The various systemd-udevd "ethtool" buffer settings now understand
1027 the special value "max" to configure the buffers to the maximum the
1030 * systemd-udevd's .link files may now configure a large variety of
1031 NIC coalescing settings, plus more hardware offload settings.
1033 * .link files gained a new WakeOnLanPassword= setting in the [Link]
1034 section that allows to specify a WoL "SecureOn" password on hardware
1037 * systemd-nspawn's --setenv= switch now supports an additional syntax:
1038 if only a variable name is specified (i.e. without being suffixed by
1039 a '=' character and a value) the current value of the environment
1040 variable is propagated to the container. e.g. --setenv=FOO will
1041 lookup the current value of $FOO in the environment, and pass it down
1042 to the container. Similar behavior has been added to homectl's,
1043 machinectl's and systemd-run's --setenv= switch.
1045 * systemd-nspawn gained a new switch --suppress-sync= which may be used
1046 to optionally suppress the effect of the sync()/fsync()/fdatasync()
1047 system calls for the container payload. This is useful for build
1048 system environments where safety against abnormal system shutdown is
1049 not essential as all build artifacts can be regenerated any time, but
1050 the performance win is beneficial.
1052 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
1053 same value that PID 1 uses for most forked off processes.
1055 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
1056 uidmap/nouidmap options as last parameter. If "uidmap" is used the
1057 bind mounts are created with UID mapping taking place that ensures
1058 the host's file ownerships are mapped 1:1 to container file
1059 ownerships, even if user namespacing is used. This way
1060 files/directories bound into containers will no longer show up as
1061 owned by the nobody user as they typically did if no special care was
1062 taken to shift them manually.
1064 * When discovering Windows installations sd-boot will now attempt to
1065 show the Windows version.
1067 * The color scheme to use in sd-boot may now be configured at
1070 * sd-boot gained the ability to change screen resolution during
1071 boot-time, by hitting the "r" key. This will cycle through available
1072 resolutions and save the last selection.
1074 * sd-boot learnt a new hotkey "f". When pressed the system will enter
1075 firmware setup. This is useful in environments where it is difficult
1076 to hit the right keys early enough to enter the firmware, and works
1077 on any firmware regardless which key it natively uses.
1079 * sd-boot gained support for automatically booting into the menu item
1080 selected on the last boot (using the "@saved" identifier for menu
1083 * sd-boot gained support for automatically loading all EFI drivers
1084 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
1085 Partition (ESP). These drivers are loaded before the menu entries are
1086 loaded. This is useful e.g. to load additional file system drivers
1087 for the XBOOTLDR partition.
1089 * systemd-boot will now paint the input cursor on its own instead of
1090 relying on the firmware to do so, increasing compatibility with broken
1091 firmware that doesn't make the cursor reasonably visible.
1093 * sd-boot now embeds a .osrel PE section like we expect from Boot
1094 Loader Specification Type #2 Unified Kernels. This means sd-boot
1095 itself may be used in place of a Type #2 Unified Kernel. This is
1096 useful for debugging purposes as it allows chain-loading one a
1097 (development) sd-boot instance from another.
1099 * sd-boot now supports a new "devicetree" field in Boot Loader
1100 Specification Type #1 entries: if configured the specified device
1101 tree file is installed before the kernel is invoked. This is useful
1102 for installing/applying new devicetree files without updating the
1105 * Similarly, sd-stub now can read devicetree data from a PE section
1106 ".dtb" and apply it before invoking the kernel.
1108 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
1109 gained the ability to pick up credentials and sysext files, wrap them
1110 in a cpio archive, and pass as an additional initrd to the invoked
1111 Linux kernel, in effect placing those files in the /.extra/ directory
1112 of the initrd environment. This is useful to implement trusted initrd
1113 environments which are fully authenticated but still can be extended
1114 (via sysexts) and parameterized (via encrypted/authenticated
1115 credentials, see above).
1117 Credentials can be located next to the kernel image file (credentials
1118 specific to a single boot entry), or in one of the shared directories
1119 (credentials applicable to multiple boot entries).
1121 * sd-stub now comes with a full man page, that explains its feature set
1122 and how to combine a kernel image, an initrd and the stub to build a
1123 complete EFI unified kernel image, implementing Boot Loader
1124 Specification Type #2.
1126 * sd-stub may now provide the initrd to the executed kernel via the
1127 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
1128 non-x86 architectures.
1130 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
1131 may be used to set the boot menu time-out of the boot loader (for all
1132 or just the subsequent boot).
1134 * bootctl and kernel-install will now read variables
1135 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
1136 /etc/kernel/install.conf. When set, it specifies the layout to use
1137 for installation directories on the boot partition, so that tools
1138 don't need to guess it based on the already-existing directories. The
1139 only value that is defined natively is "bls", corresponding to the
1141 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
1142 kernel-install that implement a different layout can declare other
1143 values for this variable.
1145 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
1146 assumption that if the user installed sd-boot to the ESP, they intend
1147 to use the entry layout understood by sd-boot. It'll also write
1148 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
1149 (and it wasn't specified in the config file yet). Similarly,
1150 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
1151 wasn't specified in the config file yet). Effectively, those changes
1152 mean that the machine-id used for boot loader entry installation is
1153 "frozen" upon first use and becomes independent of the actual
1156 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
1157 images created for distribution ("golden images") are built with no
1158 machine-id, so that a unique machine-id can be created on the first
1159 boot. But those images may contain boot loader entries with the
1160 machine-id used during build included in paths. Using a "frozen"
1161 value allows unambiguously identifying entries that match the
1162 specific installation, while still permitting parallel installations
1165 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
1166 kernel-install to guess the installation layout. This fixes the
1167 problem where a (possibly empty) directory in the boot partition is
1168 created from a different layout causing kernel-install plugins to
1169 assume the wrong layout. A particular example of how this may happen
1170 is the grub2 package in Fedora which includes directories under /boot
1171 directly in its file list. Various other packages pull in grub2 as a
1172 dependency, so it may be installed even if unused, breaking
1173 installations that use the bls layout.
1175 * bootctl and systemd-bless-boot can now be linked statically.
1177 * systemd-sysext now optionally doesn't insist on extension-release.d/
1178 files being placed in the image under the image's file name. If the
1179 file system xattr user.extension-release.strict is set on the
1180 extension release file, it is accepted regardless of its name. This
1181 relaxes security restrictions a bit, as system extension may be
1182 attached under a wrong name this way.
1184 * udevadm's test-builtin command learnt a new --action= switch for
1185 testing the built-in with the specified action (in place of the
1188 * udevadm info gained new switches --property=/--value for showing only
1189 specific udev properties/values instead of all.
1191 * A new hwdb database has been added that contains matches for various
1192 types of signal analyzers (protocol analyzers, logic analyzers,
1193 oscilloscopes, multimeters, bench power supplies, etc.) that should
1194 be accessible to regular users.
1196 * A new hwdb database entry has been added that carries information
1197 about types of cameras (regular or infrared), and in which direction
1198 they point (front or back).
1200 * A new rule to allow console users access to rfkill by default has been
1203 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
1204 now also owned by the system group "sgx".
1206 * A new build-time meson option "extra-net-naming-schemes=" has been
1207 added to define additional naming schemes schemes for udev's network
1208 interface naming logic. This is useful for enterprise distributions
1209 and similar which want to pin the schemes of certain distribution
1210 releases under a specific name and previously had to patch the
1211 sources to introduce new named schemes.
1213 * The predictable naming logic for network interfaces has been extended
1214 to generate stable names from Xen netfront device information.
1216 * hostnamed's chassis property can now be sourced from chassis-type
1217 field encoded in devicetree (in addition to the existing DMI
1220 * systemd-cgls now optionally displays cgroup IDs and extended
1221 attributes for each cgroup. (Controllable via the new --xattr= +
1222 --cgroup-id= switches.)
1224 * coredumpctl gained a new --all switch for operating on all
1225 Journal files instead of just the local ones.
1227 * systemd-coredump will now use libdw/libelf via dlopen() rather than
1228 directly linking, allowing users to easily opt-out of backtrace/metadata
1229 analysis of core files, and reduce image sizes when this is not needed.
1231 * systemd-coredump will now analyze core files with libdw/libelf in a
1232 forked, sandboxed process.
1234 * systemd-homed will now try to unmount an activate home area in
1235 regular intervals once the user logged out fully. Previously this was
1236 attempted exactly once but if the home directory was busy for some
1237 reason it was not tried again.
1239 * systemd-homed's LUKS2 home area backend will now create a BSD file
1240 system lock on the image file while the home area is active
1241 (i.e. mounted). If a home area is found to be locked, logins are
1242 politely refused. This should improve behavior when using home areas
1243 images that are accessible via the network from multiple clients, and
1244 reduce the chance of accidental file system corruption in that case.
1246 * Optionally, systemd-homed will now drop the kernel buffer cache once
1247 a user has fully logged out, configurable via the new --drop-caches=
1250 * systemd-homed now makes use of UID mapped mounts for the home areas.
1251 If the kernel and used file system support it, files are now
1252 internally owned by the "nobody" user (i.e. the user typically used
1253 for indicating "this ownership is not mapped"), and dynamically
1254 mapped to the UID used locally on the system via the UID mapping
1255 mount logic of recent kernels. This makes migrating home areas
1256 between different systems cheaper because recursively chown()ing file
1257 system trees is no longer necessary.
1259 * systemd-homed's CIFS backend now optionally supports CIFS service
1260 names with a directory suffix, in order to place home directories in
1261 a subdirectory of a CIFS share, instead of the top-level directory.
1263 * systemd-homed's CIFS backend gained support for specifying additional
1264 mount options in the JSON user record (cifsExtraMountOptions field,
1265 and --cifs-extra-mount-options= homectl switch). This is for example
1266 useful for configuring mount options such as "noserverino" that some
1267 SMB3 services require (use that to run a homed home directory from a
1268 FritzBox SMB3 share this way).
1270 * systemd-homed will now default to btrfs' zstd compression for home
1271 areas. This is inspired by Fedora's recent decision to switch to zstd
1274 * Additional mount options to use when mounting the file system of
1275 LUKS2 volumes in systemd-homed has been added. Via the
1276 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
1277 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
1278 systemd-homed or via the luksExtraMountOptions user record JSON
1279 property. (Exposed via homectl --luks-extra-mount-options)
1281 * homectl's resize command now takes the special size specifications
1282 "min" and "max" to shrink/grow the home area to the minimum/maximum
1283 size possible, taking disk usage/space constraints and file system
1284 limitations into account. Resizing is now generally graceful: the
1285 logic will try to get as close to the specified size as possible, but
1286 not consider it a failure if the request couldn't be fulfilled
1289 * systemd-homed gained the ability to automatically shrink home areas
1290 on logout to their minimal size and grow them again on next
1291 login. This ensures that while inactive, a home area only takes up
1292 the minimal space necessary, but once activated, it provides
1293 sufficient space for the user's needs. This behavior is only
1294 supported if btrfs is used as file system inside the home area
1295 (because only for btrfs online growing/shrinking is implemented in
1296 the kernel). This behavior is now enabled by default, but may be
1297 controlled via the new --auto-resize-mode= setting of homectl.
1299 * systemd-homed gained support for automatically re-balancing free disk
1300 space among active home areas, in case the LUKS2 backends are used,
1301 and no explicit disk size was requested. This way disk space is
1302 automatically managed and home areas resized in regular intervals and
1303 manual resizing when disk space becomes scarce should not be
1304 necessary anymore. This behavior is only supported if btrfs is used
1305 within the home areas (as only then online shrinking and growing is
1306 supported), and may be configured via the new rebalanceWeight JSON
1307 user record field (as exposed via the new --rebalance-weight= homectl
1308 setting). Re-balancing is mostly automatic, but can also be requested
1309 explicitly via "homectl rebalance", which is synchronous, and thus
1310 may be used to wait until the rebalance run is complete.
1312 * userdbctl gained a --json= switch for configured the JSON formatting
1313 to use when outputting user or group records.
1315 * userdbctl gained a new --multiplexer= switch for explicitly
1316 configuring whether to use the systemd-userdbd server side user
1317 record resolution logic.
1319 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
1320 for chaining up another command to execute after completing the
1321 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
1322 configuration of a single command to invoke, this maybe used to
1323 invoke multiple: first userdbctl's own implementation, and then any
1324 other also configured in the command line.
1326 * The sd-event API gained a new function sd_event_add_inotify_fd() that
1327 is similar to sd_event_add_inotify() but accepts a file descriptor
1328 instead of a path in the file system for referencing the inode to
1331 * The sd-event API gained a new function
1332 sd_event_source_set_ratelimit_expire_callback() that may be used to
1333 define a callback function that is called whenever an event source
1334 leaves the rate limiting phase.
1336 * New documentation has been added explaining which steps are necessary
1337 to port systemd to a new architecture:
1339 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
1341 * The x-systemd.makefs option in /etc/fstab now explicitly supports
1342 ext2, ext3, and f2fs file systems.
1344 * Mount units and units generated from /etc/fstab entries with 'noauto'
1345 are now ordered the same as other units. Effectively, they will be
1346 started earlier (if something actually pulled them in) and stopped
1347 later, similarly to normal mount units that are part of
1348 fs-local.target. This change should be invisible to users, but
1349 should prevent those units from being stopped too early during
1352 * The systemd-getty-generator now honors a new kernel command line
1353 argument systemd.getty_auto= and a new environment variable
1354 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
1355 example useful to turn off gettys inside of containers or similar
1358 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
1359 (in addition to 127.0.0.53, as before). If DNS requests are sent to
1360 this address they are propagated in "bypass" mode only, i.e. are
1361 almost not processed locally, but mostly forwarded as-is to the
1362 current upstream DNS servers. This provides a stable DNS server
1363 address that proxies all requests dynamically to the right upstream
1364 DNS servers even if these dynamically change. This stub does not do
1365 mDNS/LLMNR resolution. However, it will translate look-ups to
1366 DNS-over-TLS if necessary. This new stub is particularly useful in
1367 container/VM environments, or for tethering setups: use DNAT to
1368 redirect traffic to any IP address to this stub.
1370 * systemd-importd now honors new environment variables
1371 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
1372 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
1373 generation, btrfs quota setup and disk synchronization.
1375 * systemd-importd and systemd-resolved can now be optionally built with
1376 OpenSSL instead of libgcrypt.
1378 * systemd-repart no longer requires OpenSSL.
1380 * systemd-sysusers will no longer create the redundant 'nobody' group
1381 by default, as the 'nobody' user is already created with an
1382 appropriate primary group.
1384 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
1386 * systemctl show-environment gained support for --output=json.
1388 * pam_systemd will now first try to use the X11 abstract socket, and
1389 fallback to the socket file in /tmp/.X11-unix/ only if that does not
1392 * systemd-journald will no longer go back to volatile storage
1393 regardless of configuration when its unit is restarted.
1395 * Initial support for the LoongArch architecture has been added (system
1396 call lists, GPT partition table UUIDs, etc).
1398 * systemd-journald's own logging messages are now also logged to the
1399 journal itself when systemd-journald logs to /dev/kmsg.
1401 * systemd-journald now re-enables COW for archived journal files on
1402 filesystems that support COW. One benefit of this change is that
1403 archived journal files will now get compressed on btrfs filesystems
1404 that have compression enabled.
1406 * systemd-journald now deduplicates fields in a single log message
1407 before adding it to the journal. In archived journal files, it will
1408 also punch holes for unused parts and truncate the file as
1409 appropriate, leading to reductions in disk usage.
1411 * journalctl --verify was extended with more informative error
1414 * More of sd-journal's functions are now resistant against journal file
1417 * The shutdown command learnt a new option --show, to display the
1420 * A LICENSES/ directory is now included in the git tree. It contains a
1421 README.md file that explains the licenses used by source files in
1422 this repository. It also contains the text of all applicable
1423 licenses as they appear on spdx.org.
1425 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
1426 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
1427 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
1428 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
1429 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
1430 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
1431 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
1432 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
1433 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
1434 Christian Brauner, Christian Göttsche, Christian Wehrli,
1435 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
1436 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
1437 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
1438 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
1439 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
1440 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
1441 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
1442 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
1443 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
1444 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
1445 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
1446 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
1447 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
1448 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
1449 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
1450 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
1451 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
1452 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
1453 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
1454 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
1455 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
1456 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
1457 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
1458 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
1459 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
1460 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
1461 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
1462 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
1463 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
1464 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
1465 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
1466 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
1467 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
1468 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
1469 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
1470 Дамјан Георгиевски, наб
1472 — Warsaw, 2021-12-23
1476 * When operating on disk images via the --image= switch of various
1477 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
1478 no 'root=' parameter on the kernel command line, and multiple
1479 suitable root or /usr/ partitions exist in the image, then a simple
1480 comparison inspired by strverscmp() is done on the GPT partition
1481 label, and the newest partition is picked. This permits a simple and
1482 generic whole-file-system A/B update logic where new operating system
1483 versions are dropped into partitions whose label is then updated with
1484 a matching version identifier.
1486 * systemd-sysusers now supports querying the passwords to set for the
1487 users it creates via the "credentials" logic introduced in v247: the
1488 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
1489 credentials are consulted for the password to use (either in UNIX
1490 hashed form, or literally). By default these credentials are inherited
1491 down from PID1 (which in turn imports it from a container manager if
1492 there is one). This permits easy configuration of user passwords
1493 during first boot. Example:
1495 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
1497 Note that systemd-sysusers operates in purely additive mode: it
1498 executes no operation if the declared users already exist, and hence
1499 doesn't set any passwords as effect of the command line above if the
1500 specified root user exists already in the image. (Note that
1501 --volatile=yes ensures it doesn't, though.)
1503 * systemd-firstboot now also supports querying various system
1504 parameters via the credential subsystems. Thus, as above this may be
1505 used to initialize important system parameters on first boot of
1506 previously unprovisioned images (i.e. images with a mostly empty
1509 * PID 1 may now show both the unit name and the unit description
1510 strings in its status output during boot. This may be configured with
1511 StatusUnitFormat=combined in system.conf or
1512 systemd.status-unit-format=combined on the kernel command line.
1514 * The systemd-machine-id-setup tool now supports a --image= switch for
1515 provisioning a machine ID file into an OS disk image, similar to how
1516 --root= operates on an OS file tree. This matches the existing switch
1517 of the same name for systemd-tmpfiles, systemd-firstboot, and
1518 systemd-sysusers tools.
1520 * Similarly, systemd-repart gained support for the --image= switch too.
1521 In combination with the existing --size= option, this makes the tool
1522 particularly useful for easily growing disk images in a single
1523 invocation, following the declarative rules included in the image
1526 * systemd-repart's partition configuration files gained support for a
1527 new switch MakeDirectories= which may be used to create arbitrary
1528 directories inside file systems that are created, before registering
1529 them in the partition table. This is useful in particular for root
1530 partitions to create mount point directories for other partitions
1531 included in the image. For example, a disk image that contains a
1532 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
1533 create /home/ and /var/ as empty directories in the root file system
1534 on its creation, so that the resulting image can be mounted
1535 immediately, even in read-only mode.
1537 * systemd-repart's CopyBlocks= setting gained support for the special
1538 value "auto". If used, a suitable matching partition on the booted OS
1539 is found as source to copy blocks from. This is useful when
1540 implementing replicating installers, that are booted from one medium
1541 and then stream their own root partition onto the target medium.
1543 * systemd-repart's partition configuration files gained support for a
1544 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
1545 GPT partition flags for the created partitions: this is useful for
1546 marking newly created partitions as read-only, or as not being
1547 subject for automatic mounting from creation on.
1549 * The /etc/os-release file has been extended with two new (optional)
1550 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
1551 information for OS images that are updated comprehensively and
1552 atomically as one image. Two new specifiers %M, %A now resolve to
1553 these two fields in the various configuration options that resolve
1556 * portablectl gained a new switch --extension= for enabling portable
1557 service images with extensions that follow the extension image
1558 concept introduced with v248, and thus allows layering multiple
1559 images when setting up the root filesystem of the service.
1561 * systemd-coredump will now extract ELF build-id information from
1562 processes dumping core and include it in the coredump report.
1563 Moreover, it will look for ELF .note.package sections with
1564 distribution packaging meta-information about the crashing process.
1565 This is useful to directly embed the rpm or deb (or any other)
1566 package name and version in ELF files, making it easy to match
1567 coredump reports with the specific package for which the software was
1568 compiled. This is particularly useful on environments with ELF files
1569 from multiple vendors, different distributions and versions, as is
1570 common today in our containerized and sand-boxed world. For further
1573 https://systemd.io/COREDUMP_PACKAGE_METADATA
1575 * A new udev hardware database has been added for FireWire devices
1578 * The "net_id" built-in of udev has been updated with three
1579 backwards-incompatible changes:
1581 - PCI hotplug slot names on s390 systems are now parsed as
1582 hexadecimal numbers. They were incorrectly parsed as decimal
1583 previously, or ignored if the name was not a valid decimal
1586 - PCI onboard indices up to 65535 are allowed. Previously, numbers
1587 above 16383 were rejected. This primarily impacts s390 systems,
1588 where values up to 65535 are used.
1590 - Invalid characters in interface names are replaced with "_".
1592 The new version of the net naming scheme is "v249". The previous
1593 scheme can be selected via the "net.naming-scheme=v247" kernel
1594 command line parameter.
1596 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
1597 NULL bus object, for which they will return false. Or in other words,
1598 an unallocated bus connection is neither ready nor open.
1600 * The sd-device API acquired a new API function
1601 sd_device_get_usec_initialized() that returns the monotonic time when
1602 the udev device first appeared in the database.
1604 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
1605 sd_device_get_trigger_uuid(). The former is similar to
1606 sd_device_trigger() but returns a randomly generated UUID that is
1607 associated with the synthetic uevent generated by the call. This UUID
1608 may be read from the sd_device object a monitor eventually receives,
1609 via the sd_device_get_trigger_uuid(). This interface requires kernel
1610 4.13 or above to work, and allows tracking a synthetic uevent through
1611 the entire device management stack. The "udevadm trigger --settle"
1612 logic has been updated to make use of this concept if available to
1613 wait precisely for the uevents it generates. "udevadm trigger" also
1614 gained a new parameter --uuid that prints the UUID for each generated
1617 * sd-device also gained new APIs sd_device_new_from_ifname() and
1618 sd_device_new_from_ifindex() for allocating an sd-device object for
1619 the specified network interface. The former accepts an interface name
1620 (either a primary or an alternative name), the latter an interface
1623 * The native Journal protocol has been documented. Clients may talk
1624 this as alternative to the classic BSD syslog protocol for locally
1625 delivering log records to the Journal. The protocol has been stable
1626 for a long time and in fact been implemented already in a variety
1627 of alternative client libraries. This documentation makes the support
1630 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
1632 * A new BPFProgram= setting has been added to service files. It may be
1633 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
1634 file, or a bind mount or symlink to one. This may be used to upload
1635 and manage BPF programs externally and then hook arbitrary systemd
1638 * The "home.arpa" domain that has been officially declared as the
1639 choice for domain for local home networks per RFC 8375 has been added
1640 to the default NTA list of resolved, since DNSSEC is generally not
1641 available on private domains.
1643 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
1645 * A new ManageForeignRoutingPolicyRules= setting has been added to
1646 .network files which may be used to exclude foreign-created routing
1647 policy rules from systemd-networkd management.
1649 * systemd-network-wait-online gained two new switches -4 and -6 that
1650 may be used to tweak whether to wait for only IPv4 or only IPv6
1653 * .network files gained a new RequiredFamilyForOnline= setting to
1654 fine-tune whether to require an IPv4 or IPv6 address in order to
1655 consider an interface "online".
1657 * networkctl will now show an over-all "online" state in the per-link
1660 * In .network files a new OutgoingInterface= setting has been added to
1661 specify the output interface in bridge FDB setups.
1663 * In .network files the Multipath group ID may now be configured for
1664 [NextHop] entries, via the new Group= setting.
1666 * The DHCP server logic configured in .network files gained a new
1667 setting RelayTarget= that turns the server into a DHCP server relay.
1668 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
1669 to further tweak the DHCP relay behaviour.
1671 * The DHCP server logic also gained a new ServerAddress= setting in
1672 .network files that explicitly specifies the server IP address to
1673 use. If not specified, the address is determined automatically, as
1676 * The DHCP server logic in systemd-networkd gained support for static
1677 DHCP leases, configurable via the [DHCPServerStaticLease]
1678 section. This allows explicitly mapping specific MAC addresses to
1679 fixed IP addresses and vice versa.
1681 * The RestrictAddressFamilies= setting in service files now supports a
1682 new special value "none". If specified sockets of all address
1683 families will be made unavailable to services configured that way.
1685 * systemd-fstab-generator and systemd-repart have been updated to
1686 support booting from disks that carry only a /usr/ partition but no
1687 root partition yet, and where systemd-repart can add it in on the
1688 first boot. This is useful for implementing systems that ship with a
1689 single /usr/ file system, and whose root file system shall be set up
1690 and formatted on a LUKS-encrypted volume whose key is generated
1691 locally (and possibly enrolled in the TPM) during the first boot.
1693 * The [Address] section of .network files now accepts a new
1694 RouteMetric= setting that configures the routing metric to use for
1695 the prefix route created as effect of the address configuration.
1696 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
1697 gained matching settings for their prefix routes. (The option of the
1698 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
1699 it conceptually belongs there; the old option is still understood for
1702 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
1705 * A new udev property ID_NET_DHCP_BROADCAST on network interface
1706 devices is now honoured by systemd-networkd, controlling whether to
1707 issue DHCP offers via broadcasting. This is used to ensure that s390
1708 layer 3 network interfaces work out-of-the-box with systemd-networkd.
1710 * nss-myhostname and systemd-resolved will now synthesize address
1711 records for a new special hostname "_outbound". The name will always
1712 resolve to the local IP addresses most likely used for outbound
1713 connections towards the default routes. On multi-homed hosts this is
1714 useful to have a stable handle referring to "the" local IP address
1715 that matters most, to the point where this is defined.
1717 * The Discoverable Partition Specification has been updated with a new
1718 GPT partition flag "grow-file-system" defined for its partition
1719 types. Whenever partitions with this flag set are automatically
1720 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
1721 of systemd-nspawn or other tools; and as opposed to explicit mounting
1722 via /etc/fstab), the file system within the partition is
1723 automatically grown to the full size of the partition. If the file
1724 system size already matches the partition size this flag has no
1725 effect. Previously, this functionality has been available via the
1726 explicit x-systemd.growfs mount option, and this new flag extends
1727 this to automatically discovered mounts. A new GrowFileSystem=
1728 setting has been added to systemd-repart drop-in files that allows
1729 configuring this partition flag. This new flag defaults to on for
1730 partitions automatically created by systemd-repart, except if they
1731 are marked read-only. See the specification for further details:
1733 https://systemd.io/DISCOVERABLE_PARTITIONS
1735 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
1736 section. If enabled (which is the default), and an NTP server address
1737 is acquired through a DHCP lease on this interface an explicit route
1738 to this address is created on this interface to ensure that NTP
1739 traffic to the NTP server acquired on an interface is also routed
1740 through that interface. The pre-existing RoutesToDNS= setting that
1741 implements the same for DNS servers is now enabled by default.
1743 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
1744 been added that may be used to restrict the network interfaces
1745 sockets created by the service may be bound to. This is implemented
1748 * A new ConditionFirmware= setting has been added to unit files to
1749 conditionalize on certain firmware features. At the moment it may
1750 check whether running on an UEFI system, a device.tree system, or if
1751 the system is compatible with some specified device-tree feature.
1753 * A new ConditionOSRelease= setting has been added to unit files to
1754 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
1755 operators may be used to check if some field has some specific value
1756 or do an alphanumerical comparison. Equality comparisons are useful
1757 for fields like ID, but relative comparisons for fields like
1758 VERSION_ID or IMAGE_VERSION.
1760 * hostnamed gained a new Describe() D-Bus method that returns a JSON
1761 serialization of the host data it exposes. This is exposed via
1762 "hostnamectl --json=" to acquire a host identity description in JSON.
1763 It's our intention to add a similar features to most services and
1764 objects systemd manages, in order to simplify integration with
1765 program code that can consume JSON.
1767 * Similarly, networkd gained a Describe() method on its Manager and
1768 Link bus objects. This is exposed via "networkctl --json=".
1770 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
1771 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
1772 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
1773 that is used both to get the value (when no argument is given), and
1774 to set the value (when an argument is specified). The old names
1775 continue to be supported for compatibility.
1777 * systemd-detect-virt and ConditionVirtualization= are now able to
1778 correctly identify Amazon EC2 environments.
1780 * The LogLevelMax= setting of unit files now applies not only to log
1781 messages generated *by* the service, but also to log messages
1782 generated *about* the service by PID 1. To suppress logs concerning a
1783 specific service comprehensively, set this option to a high log
1786 * bootctl gained support for a new --make-machine-id-directory= switch
1787 that allows precise control on whether to create the top-level
1788 per-machine directory in the boot partition that typically contains
1789 Type 1 boot loader entries.
1791 * During build SBAT data to include in the systemd-boot EFI PE binaries
1792 may be specified now.
1794 * /etc/crypttab learnt a new option "headless". If specified any
1795 requests to query the user interactively for passwords or PINs will
1796 be skipped. This is useful on systems that are headless, i.e. where
1797 an interactive user is generally not present.
1799 * /etc/crypttab also learnt a new option "password-echo=" that allows
1800 configuring whether the encryption password prompt shall echo the
1801 typed password and if so, do so literally or via asterisks. (The
1802 default is the same behaviour as before: provide echo feedback via
1805 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
1806 systemd-homed has been updated to allow explicit configuration of the
1807 "user presence" and "user verification" checks, as well as whether a
1808 PIN is required for authentication, via the new switches
1809 --fido2-with-user-presence=, --fido2-with-user-verification=,
1810 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
1811 features are available, and may be enabled or disabled depends on the
1814 * systemd-nspawn's --private-user= switch now accepts the special value
1815 "identity" which configures a user namespacing environment with an
1816 identity mapping of 65535 UIDs. This means the container UID 0 is
1817 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
1818 this doesn't appear to be useful, however it does reduce the attack
1819 surface a bit, since the resulting container will possess process
1820 capabilities only within its namespace and not on the host.
1822 * systemd-nspawn's --private-user-chown switch has been replaced by a
1823 more generic --private-user-ownership= switch that accepts one of
1824 three values: "chown" is equivalent to the old --private-user-chown,
1825 and "off" is equivalent to the absence of the old switch. The value
1826 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
1827 of files and directories of the underlying image to the chosen UID
1828 range for the container. "auto" is equivalent to "map" if UID mapping
1829 mount are supported, otherwise it is equivalent to "chown". The short
1830 -U switch systemd-nspawn now implies --private-user-ownership=auto
1831 instead of the old --private-user-chown. Effectively this means: if
1832 the backing file system supports UID mapping mounts the feature is
1833 now used by default if -U is used. Generally, it's a good idea to use
1834 UID mapping mounts instead of recursive chown()ing, since it allows
1835 running containers off immutable images (since no modifications of
1836 the images need to take place), and share images between multiple
1837 instances. Moreover, the recursive chown()ing operation is slow and
1838 can be avoided. Conceptually it's also a good thing if transient UID
1839 range uses do not leak into persistent file ownership anymore. TLDR:
1840 finally, the last major drawback of user namespacing has been
1841 removed, and -U should always be used (unless you use btrfs, where
1842 UID mapped mounts do not exist; or your container actually needs
1843 privileges on the host).
1845 * nss-systemd now synthesizes user and group shadow records in addition
1846 to the main user and group records. Thus, hashed passwords managed by
1847 systemd-homed are now accessible via the shadow database.
1849 * The userdb logic (and thus nss-systemd, and so on) now read
1850 additional user/group definitions in JSON format from the drop-in
1851 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
1852 /usr/lib/userdb/. This is a simple and powerful mechanism for making
1853 additional users available to the system, with full integration into
1854 NSS including the shadow databases. Since the full JSON user/group
1855 record format is supported this may also be used to define users with
1856 resource management settings and other runtime settings that
1857 pam_systemd and systemd-logind enforce at login.
1859 * The userdbctl tool gained two new switches --with-dropin= and
1860 --with-varlink= which can be used to fine-tune the sources used for
1861 user database lookups.
1863 * systemd-nspawn gained a new switch --bind-user= for binding a host
1864 user account into the container. This does three things: the user's
1865 home directory is bind mounted from the host into the container,
1866 below the /run/userdb/home/ hierarchy. A free UID is picked in the
1867 container, and a user namespacing UID mapping to the host user's UID
1868 installed. And finally, a minimal JSON user and group record (along
1869 with its hashed password) is dropped into /run/host/userdb/. These
1870 records are picked up automatically by the userdb drop-in logic
1871 describe above, and allow the user to login with the same password as
1872 on the host. Effectively this means: if host and container run new
1873 enough systemd versions making a host user available to the container
1874 is trivially simple.
1876 * systemd-journal-gatewayd now supports the switches --user, --system,
1877 --merge, --file= that are equivalent to the same switches of
1878 journalctl, and permit exposing only the specified subset of the
1881 * The OnFailure= dependency between units is now augmented with a
1882 implicit reverse dependency OnFailureOf= (this new dependency cannot
1883 be configured directly it's only created as effect of an OnFailure=
1884 dependency in the reverse order — it's visible in "systemctl show"
1885 however). Similar, Slice= now has an reverse dependency SliceOf=,
1886 that is also not configurable directly, but useful to determine all
1887 units that are members of a slice.
1889 * A pair of new dependency types between units PropagatesStopTo= +
1890 StopPropagatedFrom= has been added, that allows propagation of unit
1891 stop events between two units. It operates similar to the existing
1892 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
1894 * A new dependency type OnSuccess= has been added (plus the reverse
1895 dependency OnSuccessOf=, which cannot be configured directly, but
1896 exists only as effect of the reverse OnSuccess=). It is similar to
1897 OnFailure=, but triggers in the opposite case: when a service exits
1898 cleanly. This allows "chaining up" of services where one or more
1899 services are started once another service has successfully completed.
1901 * A new dependency type Upholds= has been added (plus the reverse
1902 dependency UpheldBy=, which cannot be configured directly, but exists
1903 only as effect of Upholds=). This dependency type is a stronger form
1904 of Wants=: if a unit has an UpHolds= dependency on some other unit
1905 and the former is active then the latter is started whenever it is
1906 found inactive (and no job is queued for it). This is an alternative
1907 to Restart= inside service units, but less configurable, and the
1908 request to uphold a unit is not encoded in the unit itself but in
1909 another unit that intends to uphold it.
1911 * The systemd-ask-password tool now also supports reading passwords
1912 from the credentials subsystem, via the new --credential= switch.
1914 * The systemd-ask-password tool learnt a new switch --emoji= which may
1915 be used to explicit control whether the lock and key emoji (🔐) is
1916 shown in the password prompt on suitable TTYs.
1918 * The --echo switch of systemd-ask-password now optionally takes a
1919 parameter that controls character echo. It may either show asterisks
1920 (default, as before), turn echo off entirely, or echo the typed
1921 characters literally.
1923 * The systemd-ask-password tool also gained a new -n switch for
1924 suppressing output of a trailing newline character when writing the
1925 acquired password to standard output, similar to /bin/echo's -n
1928 * New documentation has been added that describes the organization of
1929 the systemd source code tree:
1931 https://systemd.io/ARCHITECTURE
1933 * Units using ConditionNeedsUpdate= will no longer be activated in
1936 * It is now possible to list a template unit in the WantedBy= or
1937 RequiredBy= settings of the [Install] section of another template
1938 unit, which will be instantiated using the same instance name.
1940 * A new MemoryAvailable property is available for units. If the unit,
1941 or the slice(s) it is part of, have a memory limit set via MemoryMax=/
1942 MemoryHigh=, MemoryAvailable will indicate how much more memory the
1943 unit can claim before hitting the limit(s).
1945 * systemd-coredump will now try to stay below the cgroup memory limit
1946 placed on itself or one of the slices it runs under, if the storage
1947 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
1948 since files written on such filesystems count toward the cgroup memory
1949 limit. If there is not enough available memory in such cases to store
1950 the core file uncompressed, systemd-coredump will skip to compressed
1951 storage directly (if enabled) and it will avoid analyzing the core file
1952 to print backtrace and metadata in the journal.
1954 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
1955 of a path matches the configured expectations, and remove it if not.
1957 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
1958 specify which of the several available filesystem timestamps (access
1959 time, birth time, change time, modification time) to look at when
1960 deciding whether a path has aged enough to be cleaned.
1962 * A new IPv6StableSecretAddress= setting has been added to .network
1963 files, which takes an IPv6 address to use as secret for IPv6 address
1966 * The [DHCPServer] logic in .network files gained support for a new
1967 UplinkInterface= setting that permits configuration of the uplink
1968 interface name to propagate DHCP lease information from.
1970 * The WakeOnLan= setting in .link files now accepts a list of flags
1971 instead of a single one, to configure multiple wake-on-LAN policies.
1973 * User-space defined tracepoints (USDT) have been added to udev at
1974 strategic locations. This is useful for tracing udev behaviour and
1975 performance with bpftrace and similar tools.
1977 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
1978 setting a network timeout time.
1980 * If a system service is running in a new mount namespace (RootDirectory=
1981 and friends), all file systems will be mounted with MS_NOSUID by
1982 default, unless the system is running with SELinux enabled.
1984 * When enumerating time zones the timedatectl tool will now consult the
1985 'tzdata.zi' file shipped by the IANA time zone database package, in
1986 addition to 'zone1970.tab', as before. This makes sure time zone
1987 aliases are now correctly supported. Some distributions so far did
1988 not install this additional file, most do however. If you
1989 distribution does not install it yet, it might make sense to change
1992 * Intel HID rfkill event is no longer masked, since it's the only
1993 source of rfkill event on newer HP laptops. To have both backward and
1994 forward compatibility, userspace daemon needs to debounce duplicated
1995 events in a short time window.
1997 Contributions from: Aakash Singh, adrian5, Albert Brox,
1998 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
1999 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
2000 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
2001 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
2002 Christian Hesse, Daniel Schaefer, Dan Streetman,
2003 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
2004 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
2005 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
2006 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
2007 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
2008 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
2009 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
2010 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
2011 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
2012 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
2013 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
2014 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
2015 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
2016 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
2017 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
2018 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
2019 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
2020 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
2021 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
2022 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
2023 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
2024 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
2025 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
2026 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
2027 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
2029 — Edinburgh, 2021-07-07
2033 * A concept of system extension images is introduced. Such images may
2034 be used to extend the /usr/ and /opt/ directory hierarchies at
2035 runtime with additional files (even if the file system is read-only).
2036 When a system extension image is activated, its /usr/ and /opt/
2037 hierarchies and os-release information are combined via overlayfs
2038 with the file system hierarchy of the host OS.
2040 A new systemd-sysext tool can be used to merge, unmerge, list, and
2041 refresh system extension hierarchies. See
2042 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
2044 The systemd-sysext.service automatically merges installed system
2045 extensions during boot (before basic.target, but not in very early
2046 boot, since various file systems have to be mounted first).
2048 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
2049 supported system extension level.
2051 * A new ExtensionImages= unit setting can be used to apply the same
2052 system extension image concept from systemd-sysext to the namespaced
2053 file hierarchy of specific services, following the same rules and
2056 * Support for a new special "root=tmpfs" kernel command-line option has
2057 been added. When specified, a tmpfs is mounted on /, and mount.usr=
2058 should be used to point to the operating system implementation.
2060 * A new configuration file /etc/veritytab may be used to configure
2061 dm-verity integrity protection for block devices. Each line is in the
2062 format "volume-name data-device hash-device roothash options",
2063 similar to /etc/crypttab.
2065 * A new kernel command-line option systemd.verity.root_options= may be
2066 used to configure dm-verity behaviour for the root device.
2068 * The key file specified in /etc/crypttab (the third field) may now
2069 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
2070 acquired by connecting to that socket and reading from it. This
2071 allows the implementation of a service to provide key information
2072 dynamically, at the moment when it is needed.
2074 * When the hostname is set explicitly to "localhost", systemd-hostnamed
2075 will respect this. Previously such a setting would be mostly silently
2076 ignored. The goal is to honour configuration as specified by the
2079 * The fallback hostname that will be used by the system manager and
2080 systemd-hostnamed can now be configured in two new ways: by setting
2081 DEFAULT_HOSTNAME= in os-release(5), or by setting
2082 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
2083 also be configured during compilation. The environment variable is
2084 intended for testing and local overrides, the os-release(5) field is
2085 intended to allow customization by different variants of a
2086 distribution that share the same compiled packages.
2088 * The environment block of the manager itself may be configured through
2089 a new ManagerEnvironment= setting in system.conf or user.conf. This
2090 complements existing ways to set the environment block (the kernel
2091 command line for the system manager, the inherited environment and
2092 user@.service unit file settings for the user manager).
2094 * systemd-hostnamed now exports the default hostname and the source of
2095 the configured hostname ("static", "transient", or "default") as
2098 * systemd-hostnamed now exports the "HardwareVendor" and
2099 "HardwareModel" D-Bus properties, which are supposed to contain a
2100 pair of cleaned up, human readable strings describing the system's
2101 vendor and model. It's typically sourced from the firmware's DMI
2102 tables, but may be augmented from a new hwdb database. hostnamectl
2103 shows this in the status output.
2105 * Support has been added to systemd-cryptsetup for extracting the
2106 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
2107 metadata header. This allows the information how to open the
2108 encrypted device to be embedded directly in the device and obviates
2109 the need for configuration in an external file.
2111 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
2112 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
2113 pre-existing support for PKCS#11 security tokens).
2115 * systemd-repart may enroll encrypted partitions using TPM2
2116 hardware. This may be useful for example to create an encrypted /var
2117 partition bound to the machine on first boot.
2119 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
2120 and PKCS#11 security tokens to LUKS volumes, list and destroy
2123 http://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
2125 It also supports enrolling "recovery keys" and regular passphrases.
2127 * The libfido2 dependency is now based on dlopen(), so that the library
2128 is used at runtime when installed, but is not a hard runtime
2131 * systemd-cryptsetup gained support for two new options in
2132 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
2133 request synchronous processing of encryption/decryption IO.
2135 * The manager may be configured at compile time to use the fexecve()
2136 instead of the execve() system call when spawning processes. Using
2137 fexecve() closes a window between checking the security context of an
2138 executable and spawning it, but unfortunately the kernel displays
2139 stale information in the process' "comm" field, which impacts ps
2142 * The configuration option -Dcompat-gateway-hostname has been dropped.
2143 "_gateway" is now the only supported name.
2145 * The ConditionSecurity=tpm2 unit file setting may be used to check if
2146 the system has at least one TPM2 (tpmrm class) device.
2148 * A new ConditionCPUFeature= has been added that may be used to
2149 conditionalize units based on CPU features. For example,
2150 ConditionCPUFeature=rdrand will condition a unit so that it is only
2151 run when the system CPU supports the RDRAND opcode.
2153 * The existing ConditionControlGroupController= setting has been
2154 extended with two new values "v1" and "v2". "v2" means that the
2155 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
2156 hierarchy or the hybrid hierarchy are used.
2158 * A new PrivateIPC= setting on a unit file allows executed processes to
2159 be moved into a private IPC namespace, with separate System V IPC
2160 identifiers and POSIX message queues.
2162 A new IPCNamespacePath= allows the unit to be joined to an existing
2165 * The tables of system calls in seccomp filters are now automatically
2166 generated from kernel lists exported on
2167 https://fedora.juszkiewicz.com.pl/syscalls.html.
2169 The following architectures should now have complete lists:
2170 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
2171 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
2173 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
2174 on /run/ if it is not already a mount point. A writable /run/ has
2175 always been a requirement for a functioning system, but this was not
2176 guaranteed when using a read-only image.
2178 Users can always specify BindPaths= or InaccessiblePaths= as
2179 overrides, and they will take precedence. If the host's root mount
2180 point is used, there is no change in behaviour.
2182 * New bind mounts and file system image mounts may be injected into the
2183 mount namespace of a service (without restarting it). This is exposed
2184 respectively as 'systemctl bind <unit> <path>…' and
2185 'systemctl mount-image <unit> <image>…'.
2187 * The StandardOutput= and StandardError= settings can now specify files
2188 to be truncated for output (as "truncate:<path>").
2190 * The ExecPaths= and NoExecPaths= settings may be used to specify
2191 noexec for parts of the file system.
2193 * sd-bus has a new function sd_bus_open_user_machine() to open a
2194 connection to the session bus of a specific user in a local container
2195 or on the local host. This is exposed in the existing -M switch to
2196 systemctl and similar tools:
2198 systemctl --user -M lennart@foobar start foo
2200 This will connect to the user bus of a user "lennart" in container
2201 "foobar". If no container name is specified, the specified user on
2202 the host itself is connected to
2204 systemctl --user -M lennart@ start quux
2206 * sd-bus also gained a convenience function sd_bus_message_send() to
2207 simplify invocations of sd_bus_send(), taking only a single
2208 parameter: the message to send.
2210 * sd-event allows rate limits to be set on event sources, for dealing
2211 with high-priority event sources that might starve out others. See
2212 the new man page sd_event_source_set_ratelimit(3) for details.
2214 * systemd.link files gained a [Link] Promiscuous= switch, which allows
2215 the device to be raised in promiscuous mode.
2217 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
2218 number of TX and RX queues to be configured.
2220 New [Link] TransmitQueueLength= setting allows the size of the TX
2221 queue to be configured.
2223 New [Link] GenericSegmentOffloadMaxBytes= and
2224 GenericSegmentOffloadMaxSegments= allow capping the packet size and
2225 the number of segments accepted in Generic Segment Offload.
2227 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
2228 wireless routing protocol that operates on ISO/OSI Layer 2 only and
2229 uses ethernet frames to route/bridge packets. This encompasses a new
2230 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
2231 new settings in .netdev files, and a new BatmanAdvanced= setting in
2234 * systemd.network files gained a [Network] RouteTable= configuration
2235 switch to select the routing policy table.
2237 systemd.network files gained a [RoutingPolicyRule] Type=
2238 configuration switch (one of "blackhole, "unreachable", "prohibit").
2240 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
2241 RouteAllowList= settings to ignore/accept route advertisements from
2242 routers matching specified prefixes. The DenyList= setting has been
2243 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
2246 systemd.network files gained a [DHCPv6] UseAddress= setting to
2247 optionally ignore the address provided in the lease.
2249 systemd.network files gained a [DHCPv6PrefixDelegation]
2250 ManageTemporaryAddress= switch.
2252 systemd.network files gained a new ActivationPolicy= setting which
2253 allows configuring how the UP state of an interface shall be managed,
2254 i.e. whether the interface is always upped, always downed, or may be
2255 upped/downed by the user using "ip link set dev".
2257 * The default for the Broadcast= setting in .network files has slightly
2258 changed: the broadcast address will not be configured for wireguard
2261 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
2262 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
2263 configuration options for VLAN packet handling.
2265 * udev rules may now set log_level= option. This allows debug logs to
2266 be enabled for select events, e.g. just for a specific subsystem or
2267 even a single device.
2269 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
2270 DATA_PREPARED_ID properties for block devices with ISO9660 file
2273 * udev now exports decoded DMI information about installed memory slots
2274 as device properties under the /sys/class/dmi/id/ pseudo device.
2276 * /dev/ is not mounted noexec anymore. This didn't provide any
2277 significant security benefits and would conflict with the executable
2278 mappings used with /dev/sgx device nodes. The previous behaviour can
2279 be restored for individual services with NoExecPaths=/dev (or by allow-
2280 listing and excluding /dev from ExecPaths=).
2282 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
2283 and /dev/vhost-net are owned by the kvm group.
2285 * The hardware database has been extended with a list of fingerprint
2286 readers that correctly support USB auto-suspend using data from
2289 * systemd-resolved can now answer DNSSEC questions through the stub
2290 resolver interface in a way that allows local clients to do DNSSEC
2291 validation themselves. For a question with DO+CD set, it'll proxy the
2292 DNS query and respond with a mostly unmodified packet received from
2293 the upstream server.
2295 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
2296 resolved.conf. If true the service will provide caching even for DNS
2297 lookups made to an upstream DNS server on the 127.0.0.1/::1
2298 addresses. By default (and when the option is false) systemd-resolved
2299 will not cache such lookups, in order to avoid duplicate local
2300 caching, under the assumption the local upstream server caches
2303 * systemd-resolved now implements RFC5001 NSID in its local DNS
2304 stub. This may be used by local clients to determine whether they are
2305 talking to the DNS resolver stub or a different DNS server.
2307 * When resolving host names and other records resolvectl will now
2308 report where the data was acquired from (i.e. the local cache, the
2309 network, locally synthesized, …) and whether the network traffic it
2310 effected was encrypted or not. Moreover the tool acquired a number of
2311 new options --cache=, --synthesize=, --network=, --zone=,
2312 --trust-anchor=, --validate= that take booleans and may be used to
2313 tweak a lookup, i.e. whether it may be answered from cached
2314 information, locally synthesized information, information acquired
2315 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
2316 anchor, and whether DNSSEC validation shall be executed for the
2319 * systemd-nspawn gained a new --ambient-capability= setting
2320 (AmbientCapability= in .nspawn files) to configure ambient
2321 capabilities passed to the container payload.
2323 * systemd-nspawn gained the ability to configure the firewall using the
2324 nftables subsystem (in addition to the existing iptables
2325 support). Similarly, systemd-networkd's IPMasquerade= option now
2326 supports nftables as back-end, too. In both cases NAT on IPv6 is now
2327 supported too, in addition to IPv4 (the iptables back-end still is
2330 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
2331 retains its meaning, but has been deprecated. Please switch to either
2332 "ivp4" or "both" (if covering IPv6 is desired).
2334 * systemd-importd will now download .verity and .roothash.p7s files
2335 along with the machine image (as exposed via machinectl pull-raw).
2337 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
2338 setting to configure the time a unit's cgroup needs to exceed memory
2339 pressure limits before action will be taken, and a new
2340 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
2343 systemd-oomd is now considered fully supported (the usual
2344 backwards-compatiblity promises apply). Swap is not required for
2345 operation, but it is still recommended.
2347 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
2348 configures the retry delay when trying to contact servers.
2350 * systemd-stdio-bridge gained --system/--user options to connect to the
2351 system bus (previous default) or the user session bus.
2353 * systemd-localed may now call locale-gen to generate missing locales
2354 on-demand (UTF-8-only). This improves integration with Debian-based
2355 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
2357 * systemctl --check-inhibitors=true may now be used to obey inhibitors
2358 even when invoked non-interactively. The old --ignore-inhibitors
2359 switch is now deprecated and replaced by --check-inhibitors=false.
2361 * systemctl import-environment will now emit a warning when called
2362 without any arguments (i.e. to import the full environment block of
2363 the called program). This command will usually be invoked from a
2364 shell, which means that it'll inherit a bunch of variables which are
2365 specific to that shell, and usually to the TTY the shell is connected
2366 to, and don't have any meaning in the global context of the system or
2367 user service manager. Instead, only specific variables should be
2368 imported into the manager environment block.
2370 Similarly, programs which update the manager environment block by
2371 directly calling the D-Bus API of the manager, should also push
2372 specific variables, and not the full inherited environment.
2374 * systemctl's status output now shows unit state with a more careful
2375 choice of Unicode characters: units in maintenance show a "○" symbol
2376 instead of the usual "●", failed units show "×", and services being
2379 * coredumpctl gained a --debugger-arguments= switch to pass arguments
2380 to the debugger. It also gained support for showing coredump info in
2381 a simple JSON format.
2383 * systemctl/loginctl/machinectl's --signal= option now accept a special
2384 value "list", which may be used to show a brief table with known
2385 process signals and their numbers.
2387 * networkctl now shows the link activation policy in status.
2389 * Various tools gained --pager/--no-pager/--json= switches to
2390 enable/disable the pager and provide JSON output.
2392 * Various tools now accept two new values for the SYSTEMD_COLORS
2393 environment variable: "16" and "256", to configure how many terminal
2394 colors are used in output.
2396 * less 568 or newer is now required for the auto-paging logic of the
2397 various tools. Hyperlink ANSI sequences in terminal output are now
2398 used even if a pager is used, and older versions of less are not able
2399 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
2400 disable this output again.
2402 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
2403 builds, non-merged-usr builds) are now officially deprecated. A
2404 warning is emitted during build. Support is slated to be removed in
2405 about a year (when the Debian Bookworm release development starts).
2407 * Systems with the legacy cgroup v1 hierarchy are now marked as
2408 "tainted", to make it clearer that using the legacy hierarchy is not
2411 * systemd-localed will now refuse to configure a keymap which is not
2412 installed in the file system. This is intended as a bug fix, but
2413 could break cases where systemd-localed was used to configure the
2414 keymap in advanced of it being installed. It is necessary to install
2415 the keymap file first.
2417 * The main git development branch has been renamed to 'main'.
2419 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
2420 for partitions, as in the vast majority of cases they contain none
2421 and are used internally by the bootloader (eg: uboot).
2423 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
2424 spawned processes to the PID of the process itself. This may be used
2425 by programs for detecting whether they were forked off by the service
2426 manager itself or are a process forked off further down the tree.
2428 * The sd-device API gained four new calls: sd_device_get_action() to
2429 determine the uevent add/remove/change/… action the device object has
2430 been seen for, sd_device_get_seqno() to determine the uevent sequence
2431 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
2432 object from stat(2) data of a device node, and sd_device_trigger() to
2433 write to the 'uevent' attribute of a device.
2435 * For most tools the --no-legend= switch has been replaced by
2436 --legend=no and --legend=yes, to force whether tables are shown with
2439 * Units acquired a new property "Markers" that takes a list of zero,
2440 one or two of the following strings: "needs-reload" and
2441 "needs-restart". These markers may be set via "systemctl
2442 set-property". Once a marker is set, "systemctl reload-or-restart
2443 --marked" may be invoked to execute the operation the units are
2444 marked for. This is useful for package managers that want to mark
2445 units for restart/reload while updating, but effect the actual
2446 operations at a later step at once.
2448 * The sd_bus_message_read_strv() API call of sd-bus may now also be
2449 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
2452 * bootctl will now report whether the UEFI firmware used a TPM2 device
2453 and measured the boot process into it.
2455 * systemd-tmpfiles learnt support for a new environment variable
2456 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
2457 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
2458 even if the root fs of the system is not itself a btrfs volume.
2460 * systemd-detect-virt/ConditionVirtualization= will now explicitly
2461 detect Docker/Podman environments where possible. Moreover, they
2462 should be able to generically detect any container manager as long as
2463 it assigns the container a cgroup.
2465 * portablectl gained a new "reattach" verb for detaching/reattaching a
2466 portable service image, useful for updating images on-the-fly.
2468 * Intel SGX enclave device nodes (which expose a security feature of
2469 newer Intel CPUs) will now be owned by a new system group "sgx".
2471 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
2472 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
2473 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
2474 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
2475 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
2476 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
2477 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
2478 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
2479 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
2480 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
2481 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
2482 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
2483 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
2484 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
2485 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
2486 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
2487 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
2488 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
2489 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
2490 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
2491 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
2492 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
2493 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
2494 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
2495 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
2496 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
2497 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
2498 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
2499 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
2500 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
2501 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
2502 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
2503 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
2504 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
2505 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
2506 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
2507 Zmicer Turok, Дамјан Георгиевски
2509 — Berlin, 2021-03-30
2513 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
2514 "bind" and "unbind" to the Linux device model. When this kernel
2515 change was made, systemd-udevd was only minimally updated to handle
2516 and propagate these new event types. The introduction of these new
2517 uevents (which are typically generated for USB devices and devices
2518 needing a firmware upload before being functional) resulted in a
2519 number of issues which we so far didn't address. We hoped the kernel
2520 maintainers would themselves address these issues in some form, but
2521 that did not happen. To handle them properly, many (if not most) udev
2522 rules files shipped in various packages need updating, and so do many
2523 programs that monitor or enumerate devices with libudev or sd-device,
2524 or otherwise process uevents. Please note that this incompatibility
2525 is not fault of systemd or udev, but caused by an incompatible kernel
2526 change that happened back in Linux 4.14, but is becoming more and
2527 more visible as the new uevents are generated by more kernel drivers.
2529 To minimize issues resulting from this kernel change (but not avoid
2530 them entirely) starting with systemd-udevd 247 the udev "tags"
2531 concept (which is a concept for marking and filtering devices during
2532 enumeration and monitoring) has been reworked: udev tags are now
2533 "sticky", meaning that once a tag is assigned to a device it will not
2534 be removed from the device again until the device itself is removed
2535 (i.e. unplugged). This makes sure that any application monitoring
2536 devices that match a specific tag is guaranteed to both see uevents
2537 where the device starts being relevant, and those where it stops
2538 being relevant (the latter now regularly happening due to the new
2539 "unbind" uevent type). The udev tags concept is hence now a concept
2540 tied to a *device* instead of a device *event* — unlike for example
2541 udev properties whose lifecycle (as before) is generally tied to a
2542 device event, meaning that the previously determined properties are
2543 forgotten whenever a new uevent is processed.
2545 With the newly redefined udev tags concept, sometimes it's necessary
2546 to determine which tags are the ones applied by the most recent
2547 uevent/database update, in order to discern them from those
2548 originating from earlier uevents/database updates of the same
2549 device. To accommodate for this a new automatic property CURRENT_TAGS
2550 has been added that works similar to the existing TAGS property but
2551 only lists tags set by the most recent uevent/database
2552 update. Similarly, the libudev/sd-device API has been updated with
2553 new functions to enumerate these 'current' tags, in addition to the
2554 existing APIs that now enumerate the 'sticky' ones.
2556 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
2557 essential that all udev rules files and applications are updated to
2558 handle the new events. Specifically:
2560 • All rule files that currently use a header guard similar to
2561 ACTION!="add|change",GOTO="xyz_end" should be updated to use
2562 ACTION=="remove",GOTO="xyz_end" instead, so that the
2563 properties/tags they add are also applied whenever "bind" (or
2564 "unbind") is seen. (This is most important for all physical device
2565 types — those for which "bind" and "unbind" are currently
2566 generated, for all other device types this change is still
2567 recommended but not as important — but certainly prepares for
2568 future kernel uevent type additions).
2570 • Similarly, all code monitoring devices that contains an 'if' branch
2571 discerning the "add" + "change" uevent actions from all other
2572 uevents actions (i.e. considering devices only relevant after "add"
2573 or "change", and irrelevant on all other events) should be reworked
2574 to instead negatively check for "remove" only (i.e. considering
2575 devices relevant after all event types, except for "remove", which
2576 invalidates the device). Note that this also means that devices
2577 should be considered relevant on "unbind", even though conceptually
2578 this — in some form — invalidates the device. Since the precise
2579 effect of "unbind" is not generically defined, devices should be
2580 considered relevant even after "unbind", however I/O errors
2581 accessing the device should then be handled gracefully.
2583 • Any code that uses device tags for deciding whether a device is
2584 relevant or not most likely needs to be updated to use the new
2585 udev_device_has_current_tag() API (or sd_device_has_current_tag()
2586 in case sd-device is used), to check whether the tag is set at the
2587 moment an uevent is seen (as opposed to the existing
2588 udev_device_has_tag() API which checks if the tag ever existed on
2589 the device, following the API concept redefinition explained
2592 We are very sorry for this breakage and the requirement to update
2593 packages using these interfaces. We'd again like to underline that
2594 this is not caused by systemd/udev changes, but result of a kernel
2597 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
2598 packages have not retriggered devices once the udev package (or any
2599 auxiliary package installing additional udev rules) is updated. We
2600 intend to work with major distributions to change this, so that
2601 "udevadm trigger -a change" is issued on such upgrades, ensuring that
2602 the updated ruleset is applied to the devices already discovered, so
2603 that (asynchronously) after the upgrade completed the udev database
2604 is consistent with the updated rule set. This means udev rules must
2605 be ready to be retriggered with a "change" action any time, and
2606 result in correct and complete udev database entries. While the
2607 majority of udev rule files known to us currently get this right,
2608 some don't. Specifically, there are udev rules files included in
2609 various packages that only set udev properties on the "add" action,
2610 but do not handle the "change" action. If a device matching those
2611 rules is retriggered with the "change" action (as is intended here)
2612 it would suddenly lose the relevant properties. This always has been
2613 problematic, but as soon as all udev devices are triggered on relevant
2614 package upgrades this will become particularly so. It is strongly
2615 recommended to fix offending rules so that they can handle a "change"
2616 action at any time, and acquire all necessary udev properties even
2617 then. Or in other words: the header guard mentioned above
2618 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
2619 this, as it makes sure rules are rerun on "change" correctly, and
2620 accumulate the correct and complete set of udev properties. udev rule
2621 definitions that cannot handle "change" events being triggered at
2622 arbitrary times should be considered buggy.
2624 * The MountAPIVFS= service file setting now defaults to on if
2625 RootImage= and RootDirectory= are used, which means that with those
2626 two settings /proc/, /sys/ and /dev/ are automatically properly set
2627 up for services. Previous behaviour may be restored by explicitly
2628 setting MountAPIVFS=off.
2630 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
2631 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
2632 latter it takes precedence over the former, similar to how most of
2633 systemd's own configuration is handled. Given that PAM stack
2634 definitions are primarily put together by OS vendors/distributions
2635 (though possibly overridden by users), this systemd release moves its
2636 own PAM stack configuration for the "systemd-user" PAM service (i.e.
2637 for the PAM session invoked by the per-user user@.service instance)
2638 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
2639 packages' vendor versions of their PAM stack definitions from
2640 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
2641 desired the location to which systemd installs its PAM stack
2642 configuration may be changed via the -Dpamconfdir Meson option.
2644 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
2645 libpwquality and libcryptsetup have been changed to be based on
2646 dlopen(): instead of regular dynamic library dependencies declared in
2647 the binary ELF headers, these libraries are now loaded on demand
2648 only, if they are available. If the libraries cannot be found the
2649 relevant operations will fail gracefully, or a suitable fallback
2650 logic is chosen. This is supposed to be useful for general purpose
2651 distributions, as it allows minimizing the list of dependencies the
2652 systemd packages pull in, permitting building of more minimal OS
2653 images, while still making use of these "weak" dependencies should
2654 they be installed. Since many package managers automatically
2655 synthesize package dependencies from ELF shared library dependencies,
2656 some additional manual packaging work has to be done now to replace
2657 those (slightly downgraded from "required" to "recommended" or
2658 whatever is conceptually suitable for the package manager). Note that
2659 this change does not alter build-time behaviour: as before the
2660 build-time dependencies have to be installed during build, even if
2661 they now are optional during runtime.
2663 * sd-event.h gained a new call sd_event_add_time_relative() for
2664 installing timers relative to the current time. This is mostly a
2665 convenience wrapper around the pre-existing sd_event_add_time() call
2666 which installs absolute timers.
2668 * sd-event event sources may now be placed in a new "exit-on-failure"
2669 mode, which may be controlled via the new
2670 sd_event_source_get_exit_on_failure() and
2671 sd_event_source_set_exit_on_failure() functions. If enabled, any
2672 failure returned by the event source handler functions will result in
2673 exiting the event loop (unlike the default behaviour of just
2674 disabling the event source but continuing with the event loop). This
2675 feature is useful to set for all event sources that define "primary"
2676 program behaviour (where failure should be fatal) in contrast to
2677 "auxiliary" behaviour (where failure should remain local).
2679 * Most event source types sd-event supports now accept a NULL handler
2680 function, in which case the event loop is exited once the event
2681 source is to be dispatched, using the userdata pointer — converted to
2682 a signed integer — as exit code of the event loop. Previously this
2683 was supported for IO and signal event sources already. Exit event
2684 sources still do not support this (simply because it makes little
2685 sense there, as the event loop is already exiting when they are
2688 * A new per-unit setting RootImageOptions= has been added which allows
2689 tweaking the mount options for any file system mounted as effect of
2690 the RootImage= setting.
2692 * Another new per-unit setting MountImages= has been added, that allows
2693 mounting additional disk images into the file system tree accessible
2696 * Timer units gained a new FixedRandomDelay= boolean setting. If
2697 enabled, the random delay configured with RandomizedDelaySec= is
2698 selected in a way that is stable on a given system (though still
2699 different for different units).
2701 * Socket units gained a new setting Timestamping= that takes "us", "ns"
2702 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
2705 * systemd-repart now generates JSON output when requested with the new
2708 * systemd-machined's OpenMachineShell() bus call will now pass
2709 additional policy metadata data fields to the PolicyKit
2710 authentication request.
2712 * systemd-tmpfiles gained a new -E switch, which is equivalent to
2713 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
2714 --exclude=/sys. It's particularly useful in combination with --root=,
2715 when operating on OS trees that do not have any of these four runtime
2716 directories mounted, as this means no files below these subtrees are
2717 created or modified, since those mount points should probably remain
2720 * systemd-tmpfiles gained a new --image= switch which is like --root=,
2721 but takes a disk image instead of a directory as argument. The
2722 specified disk image is mounted inside a temporary mount namespace
2723 and the tmpfiles.d/ drop-ins stored in the image are executed and
2724 applied to the image. systemd-sysusers similarly gained a new
2725 --image= switch, that allows the sysusers.d/ drop-ins stored in the
2726 image to be applied onto the image.
2728 * Similarly, the journalctl command also gained an --image= switch,
2729 which is a quick one-step solution to look at the log data included
2732 * journalctl's --output=cat option (which outputs the log content
2733 without any metadata, just the pure text messages) will now make use
2734 of terminal colors when run on a suitable terminal, similarly to the
2737 * JSON group records now support a "description" string that may be
2738 used to add a human-readable textual description to such groups. This
2739 is supposed to match the user's GECOS field which traditionally
2740 didn't have a counterpart for group records.
2742 * The "systemd-dissect" tool that may be used to inspect OS disk images
2743 and that was previously installed to /usr/lib/systemd/ has now been
2744 moved to /usr/bin/, reflecting its updated status of an officially
2745 supported tool with a stable interface. It gained support for a new
2746 --mkdir switch which when combined with --mount has the effect of
2747 creating the directory to mount the image to if it is missing
2748 first. It also gained two new commands --copy-from and --copy-to for
2749 copying files and directories in and out of an OS image without the
2750 need to manually mount it. It also acquired support for a new option
2751 --json= to generate JSON output when inspecting an OS image.
2753 * The cgroup2 file system is now mounted with the
2754 "memory_recursiveprot" mount option, supported since kernel 5.7. This
2755 means that the MemoryLow= and MemoryMin= unit file settings now apply
2756 recursively to whole subtrees.
2758 * systemd-homed now defaults to using the btrfs file system — if
2759 available — when creating home directories in LUKS volumes. This may
2760 be changed with the DefaultFileSystemType= setting in homed.conf.
2761 It's now the default file system in various major distributions and
2762 has the major benefit for homed that it can be grown and shrunk while
2763 mounted, unlike the other contenders ext4 and xfs, which can both be
2764 grown online, but not shrunk (in fact xfs is the technically most
2765 limited option here, as it cannot be shrunk at all).
2767 * JSON user records managed by systemd-homed gained support for
2768 "recovery keys". These are basically secondary passphrases that can
2769 unlock user accounts/home directories. They are computer-generated
2770 rather than user-chosen, and typically have greater entropy.
2771 homectl's --recovery-key= option may be used to add a recovery key to
2772 a user account. The generated recovery key is displayed as a QR code,
2773 so that it can be scanned to be kept in a safe place. This feature is
2774 particularly useful in combination with systemd-homed's support for
2775 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
2776 security tokens are lost. Recovery keys may be entered wherever the
2777 system asks for a password.
2779 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
2780 home directory which indicates that a home directory has not been
2781 deactivated cleanly when offline. This flag is useful to identify
2782 home directories for which the offline discard logic did not run when
2783 offlining, and where it would be a good idea to log in again to catch
2786 * systemctl gained a new parameter --timestamp= which may be used to
2787 change the style in which timestamps are output, i.e. whether to show
2788 them in local timezone or UTC, or whether to show µs granularity.
2790 * Alibaba's "pouch" container manager is now detected by
2791 systemd-detect-virt, ConditionVirtualization= and similar
2792 constructs. Similar, they now also recognize IBM PowerVM machine
2795 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
2796 place to use for propagating external mounts into the
2797 container. Similarly /run/host/notify is now used as the socket path
2798 for container payloads to communicate with the container manager
2799 using sd_notify(). The container manager now uses the
2800 /run/host/inaccessible/ directory to place "inaccessible" file nodes
2801 of all relevant types which may be used by the container payload as
2802 bind mount source to over-mount inodes to make them inaccessible.
2803 /run/host/container-manager will now be initialized with the same
2804 string as the $container environment variable passed to the
2805 container's PID 1. /run/host/container-uuid will be initialized with
2806 the same string as $container_uuid. This means the /run/host/
2807 hierarchy is now the primary way to make host resources available to
2808 the container. The Container Interface documents these new files and
2811 https://systemd.io/CONTAINER_INTERFACE
2813 * Support for the "ConditionNull=" unit file condition has been
2814 deprecated and undocumented for 6 years. systemd started to warn
2815 about its use 1.5 years ago. It has now been removed entirely.
2817 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
2818 a sd_bus_error struct and a list of error names, and checks if the
2819 error matches one of these names. It's a convenience wrapper that is
2820 useful in cases where multiple errors shall be handled the same way.
2822 * A new system call filter list "@known" has been added, that contains
2823 all system calls known at the time systemd was built.
2825 * Behaviour of system call filter allow lists has changed slightly:
2826 system calls that are contained in @known will result in EPERM by
2827 default, while those not contained in it result in ENOSYS. This
2828 should improve compatibility because known system calls will thus be
2829 communicated as prohibited, while unknown (and thus newer ones) will
2830 be communicated as not implemented, which hopefully has the greatest
2831 chance of triggering the right fallback code paths in client
2834 * "systemd-analyze syscall-filter" will now show two separate sections
2835 at the bottom of the output: system calls known during systemd build
2836 time but not included in any of the filter groups shown above, and
2837 system calls defined on the local kernel but known during systemd
2840 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
2841 systemd-nspawn all system call filter violations will be logged by
2842 the kernel (audit). This is useful for tracking down system calls
2843 invoked by container payloads that are prohibited by the container's
2844 system call filter policy.
2846 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
2847 systemd-nspawn (and other programs that use seccomp) all seccomp
2848 filtering is turned off.
2850 * Two new unit file settings ProtectProc= and ProcSubset= have been
2851 added that expose the hidepid= and subset= mount options of procfs.
2852 All processes of the unit will only see processes in /proc that are
2853 are owned by the unit's user. This is an important new sandboxing
2854 option that is recommended to be set on all system services. All
2855 long-running system services that are included in systemd itself set
2856 this option now. This option is only supported on kernel 5.8 and
2857 above, since the hidepid= option supported on older kernels was not a
2858 per-mount option but actually applied to the whole PID namespace.
2860 * Socket units gained a new boolean setting FlushPending=. If enabled
2861 all pending socket data/connections are flushed whenever the socket
2862 unit enters the "listening" state, i.e. after the associated service
2865 * The unit file setting NUMAMask= gained a new "all" value: when used,
2866 all existing NUMA nodes are added to the NUMA mask.
2868 * A new "credentials" logic has been added to system services. This is
2869 a simple mechanism to pass privileged data to services in a safe and
2870 secure way. It's supposed to be used to pass per-service secret data
2871 such as passwords or cryptographic keys but also associated less
2872 private information such as user names, certificates, and similar to
2873 system services. Each credential is identified by a short user-chosen
2874 name and may contain arbitrary binary data. Two new unit file
2875 settings have been added: SetCredential= and LoadCredential=. The
2876 former allows setting a credential to a literal string, the latter
2877 sets a credential to the contents of a file (or data read from a
2878 user-chosen AF_UNIX stream socket). Credentials are passed to the
2879 service via a special credentials directory, one file for each
2880 credential. The path to the credentials directory is passed in a new
2881 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
2882 are passed in the file system they may be easily referenced in
2883 ExecStart= command lines too, thus no explicit support for the
2884 credentials logic in daemons is required (though ideally daemons
2885 would look for the bits they need in $CREDENTIALS_DIRECTORY
2886 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
2887 backed by unswappable memory if privileges allow it, immutable if
2888 privileges allow it, is accessible only to the service's UID, and is
2889 automatically destroyed when the service stops.
2891 * systemd-nspawn supports the same credentials logic. It can both
2892 consume credentials passed to it via the aforementioned
2893 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
2894 to its payload. The service manager/PID 1 has been updated to match
2895 this: it can also accept credentials from the container manager that
2896 invokes it (in fact: any process that invokes it), and passes them on
2897 to its services. Thus, credentials can be propagated recursively down
2898 the tree: from a system's service manager to a systemd-nspawn
2899 service, to the service manager that runs as container payload and to
2900 the service it runs below. Credentials may also be added on the
2901 systemd-nspawn command line, using new --set-credential= and
2902 --load-credential= command line switches that match the
2903 aforementioned service settings.
2905 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
2906 the partition drop-ins which may be used to format/LUKS
2907 encrypt/populate any created partitions. The partitions are
2908 encrypted/formatted/populated before they are registered in the
2909 partition table, so that they appear atomically: either the
2910 partitions do not exist yet or they exist fully encrypted, formatted,
2911 and populated — there is no time window where they are
2912 "half-initialized". Thus the system is robust to abrupt shutdown: if
2913 the tool is terminated half-way during its operations on next boot it
2914 will start from the beginning.
2916 * systemd-repart's --size= operation gained a new "auto" value. If
2917 specified, and operating on a loopback file it is automatically sized
2918 to the minimal size the size constraints permit. This is useful to
2919 use "systemd-repart" as an image builder for minimally sized images.
2921 * systemd-resolved now gained a third IPC interface for requesting name
2922 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
2923 interface is now supported. The nss-resolve NSS module has been
2924 modified to use this new interface instead of D-Bus. Using Varlink
2925 has a major benefit over D-Bus: it works without a broker service,
2926 and thus already during earliest boot, before the dbus daemon has
2927 been started. This means name resolution via systemd-resolved now
2928 works at the same time systemd-networkd operates: from earliest boot
2929 on, including in the initrd.
2931 * systemd-resolved gained support for a new DNSStubListenerExtra=
2932 configuration file setting which may be used to specify additional IP
2933 addresses the built-in DNS stub shall listen on, in addition to the
2934 main one on 127.0.0.53:53.
2936 * Name lookups issued via systemd-resolved's D-Bus and Varlink
2937 interfaces (and thus also via glibc NSS if nss-resolve is used) will
2938 now honour a trailing dot in the hostname: if specified the search
2939 path logic is turned off. Thus "resolvectl query foo." is now
2940 equivalent to "resolvectl query --search=off foo.".
2942 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
2943 exposes how /etc/resolv.conf is currently managed: by resolved (and
2944 in which mode if so) or another subsystem. "resolvctl" will display
2945 this property in its status output.
2947 * The resolv.conf snippets systemd-resolved provides will now set "."
2948 as the search domain if no other search domain is known. This turns
2949 off the derivation of an implicit search domain by nss-dns for the
2950 hostname, when the hostname is set to an FQDN. This change is done to
2951 make nss-dns using resolv.conf provided by systemd-resolved behave
2952 more similarly to nss-resolve.
2954 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
2955 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
2956 "birth" time (btime) of a file in addition to the atime, mtime, and
2959 * systemd-analyze gained a new verb "capability" that lists all known
2960 capabilities by the systemd build and by the kernel.
2962 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
2963 advance the system clock to it at boot if it is noticed to be before
2964 that time. Previously, PID 1 would only advance the time to an epoch
2965 time that is set during build-time. With this new file OS builders
2966 can change this epoch timestamp on individual OS images without
2967 having to rebuild systemd.
2969 * systemd-logind will now listen to the KEY_RESTART key from the Linux
2970 input layer and reboot the system if it is pressed, similarly to how
2971 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
2972 was originally defined in the Multimedia context (to restart playback
2973 of a song or film), but is now primarily used in various embedded
2974 devices for "Reboot" buttons. Accordingly, systemd-logind will now
2975 honour it as such. This may configured in more detail via the new
2976 HandleRebootKey= and RebootKeyIgnoreInhibited=.
2978 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
2979 copying OS trees, for example in "systemd-nspawn --ephemeral",
2980 "systemd-nspawn --template=", "machinectl clone" and similar. This is
2981 useful when operating with OSTree images, which use hardlinks heavily
2982 throughout, and where such copies previously resulting in "exploding"
2985 * systemd-nspawn's --console= setting gained support for a new
2986 "autopipe" value, which is identical to "interactive" when invoked on
2987 a TTY, and "pipe" otherwise.
2989 * systemd-networkd's .network files gained support for explicitly
2990 configuring the multicast membership entries of bridge devices in the
2991 [BridgeMDB] section. It also gained support for the PIE queuing
2992 discipline in the [FlowQueuePIE] sections.
2994 * systemd-networkd's .netdev files may now be used to create "BareUDP"
2995 tunnels, configured in the new [BareUDP] setting.
2997 * systemd-networkd's Gateway= setting in .network files now accepts the
2998 special values "_dhcp4" and "_ipv6ra" to configure additional,
2999 locally defined, explicit routes to the gateway acquired via DHCP or
3000 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
3001 but still accepted for backwards compatibility.
3003 * systemd-networkd's [IPv6PrefixDelegation] section and
3004 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
3005 IPv6SendRA= (the old names are still accepted for backwards
3008 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
3009 boolean setting in [Network] section. If enabled, the delegated prefix
3010 gained by another link will be configured, and an address within the
3011 prefix will be assigned.
3013 * systemd-networkd's .network files gained the Announce= boolean setting
3014 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
3015 prefix will be announced through IPv6 router advertisement (IPv6 RA).
3016 The setting is enabled by default.
3018 * VXLAN tunnels may now be marked as independent of any underlying
3019 network interface via the new Independent= boolean setting.
3021 * systemctl gained support for two new verbs: "service-log-level" and
3022 "service-log-target" may be used on services that implement the
3023 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
3024 adjust the log level and target. All of systemd's long-running
3025 services support this now, but ideally all system services would
3026 implement this interface to make the system more uniformly
3029 * The SystemCallErrorNumber= unit file setting now accepts the new
3030 "kill" and "log" actions, in addition to arbitrary error number
3031 specifications as before. If "kill" the processes are killed on the
3032 event, if "log" the offending system call is audit logged.
3034 * A new SystemCallLog= unit file setting has been added that accepts a
3035 list of system calls that shall be logged about (audit).
3037 * The OS image dissection logic (as used by RootImage= in unit files or
3038 systemd-nspawn's --image= switch) has gained support for identifying
3039 and mounting explicit /usr/ partitions, which are now defined in the
3040 discoverable partition specification. This should be useful for
3041 environments where the root file system is
3042 generated/formatted/populated dynamically on first boot and combined
3043 with an immutable /usr/ tree that is supplied by the vendor.
3045 * In the final phase of shutdown, within the systemd-shutdown binary
3046 we'll now try to detach MD devices (i.e software RAID) in addition to
3047 loopback block devices and DM devices as before. This is supposed to
3048 be a safety net only, in order to increase robustness if things go
3049 wrong. Storage subsystems are expected to properly detach their
3050 storage volumes during regular shutdown already (or in case of
3051 storage backing the root file system: in the initrd hook we return to
3054 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
3055 will now log the thread ID in their log output. This is useful when
3056 working with heavily threaded programs.
3058 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
3059 not use the RDRAND CPU instruction. This is useful in environments
3060 such as replay debuggers where non-deterministic behaviour is not
3063 * The autopaging logic in systemd's various tools (such as systemctl)
3064 has been updated to turn on "secure" mode in "less"
3065 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
3066 detected. This disables invoking external programs from the pager,
3067 via the pipe logic. This behaviour may be overridden via the new
3068 $SYSTEMD_PAGERSECURE environment variable.
3070 * Units which have resource limits (.service, .mount, .swap, .slice,
3071 .socket, and .slice) gained new configuration settings
3072 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
3073 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
3074 limits and optional action taken by systemd-oomd.
3076 * A new service systemd-oomd has been added. It monitors resource
3077 contention for selected parts of the unit hierarchy using the PSI
3078 information reported by the kernel, and kills processes when memory
3079 or swap pressure is above configured limits. This service is only
3080 enabled by default in developer mode (see below) and should be
3081 considered a preview in this release. Behaviour details and option
3082 names are subject to change without the usual backwards-compatibility
3085 * A new helper oomctl has been added to introspect systemd-oomd state.
3086 It is only enabled by default in developer mode and should be
3087 considered a preview without the usual backwards-compatibility
3090 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
3091 enabled, systemd reads the system UID boundaries from /etc/login.defs
3092 at runtime, instead of using the built-in values selected during
3093 build. This is an option to improve compatibility for upgrades from
3094 old systems. It's strongly recommended not to make use of this
3095 functionality on new systems (or even enable it during build), as it
3096 makes something runtime-configurable that is mostly an implementation
3097 detail of the OS, and permits avoidable differences in deployments
3098 that create all kinds of problems in the long run.
3100 * New meson option '-Dmode=developer|release' has been added. When
3101 'developer', additional checks and features are enabled that are
3102 relevant during upstream development, e.g. verification that
3103 semi-automatically-generated documentation has been properly updated
3104 following API changes. Those checks are considered hints for
3105 developers and are not actionable in downstream builds. In addition,
3106 extra features that are not ready for general consumption may be
3107 enabled in developer mode. It is thus recommended to set
3108 '-Dmode=release' in end-user and distro builds.
3110 * systemd-cryptsetup gained support for processing detached LUKS
3111 headers specified on the kernel command line via the header=
3112 parameter of the luks.options= kernel command line option. The same
3113 device/path syntax as for key files is supported for header files
3116 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
3117 slot index data for devices that are connected through a PCI bridge
3118 where the _SUN index is associated with the bridge instead of the
3119 network device itself. Previously this would create ambiguous device
3120 naming if multiple network interfaces were connected to the same PCI
3121 bridge. Since this is a naming scheme incompatibility on systems that
3122 possess hardware like this it has been introduced as new naming
3123 scheme "v247". The previous scheme can be selected via the
3124 "net.naming-scheme=v245" kernel command line parameter.
3126 * ConditionFirstBoot= semantics have been modified to be safe towards
3127 abnormal system power-off during first boot. Specifically, the
3128 "systemd-machine-id-commit.service" service now acts as boot
3129 milestone indicating when the first boot process is sufficiently
3130 complete in order to not consider the next following boot also a
3131 first boot. If the system is reset before this unit is reached the
3132 first time, the next boot will still be considered a first boot; once
3133 it has been reached, no further boots will be considered a first
3134 boot. The "first-boot-complete.target" unit now acts as official hook
3135 point to order against this. If a service shall be run on every boot
3136 until the first boot fully succeeds it may thus be ordered before
3137 this target unit (and pull it in) and carry ConditionFirstBoot=
3140 * bootctl's set-default and set-oneshot commands now accept the three
3141 special strings "@default", "@oneshot", "@current" in place of a boot
3142 entry id. These strings are resolved to the current default and
3143 oneshot boot loader entry, as well as the currently booted one. Thus
3144 a command "bootctl set-default @current" may be used to make the
3145 currently boot menu item the new default for all subsequent boots.
3147 * "systemctl edit" has been updated to show the original effective unit
3148 contents in commented form in the text editor.
3150 * Units in user mode are now segregated into three new slices:
3151 session.slice (units that form the core of graphical session),
3152 app.slice ("normal" user applications), and background.slice
3153 (low-priority tasks). Unless otherwise configured, user units are
3154 placed in app.slice. The plan is to add resource limits and
3155 protections for the different slices in the future.
3157 * New GPT partition types for RISCV32/64 for the root and /usr
3158 partitions, and their associated Verity partitions have been defined,
3159 and are now understood by systemd-gpt-auto-generator, and the OS
3160 image dissection logic.
3162 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
3163 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
3164 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
3165 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
3166 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
3167 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
3168 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
3169 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
3170 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
3171 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
3172 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
3173 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
3174 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
3175 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
3176 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
3177 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
3178 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
3179 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
3180 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
3181 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
3182 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
3183 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
3184 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
3185 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
3186 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
3187 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
3188 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
3189 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
3190 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
3191 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
3192 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
3193 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
3194 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
3196 – Warsaw, 2020-11-26
3200 * The service manager gained basic support for cgroup v2 freezer. Units
3201 can now be suspended or resumed either using new systemctl verbs,
3202 freeze and thaw respectively, or via D-Bus.
3204 * PID 1 may now automatically load pre-compiled AppArmor policies from
3205 /etc/apparmor/earlypolicy during early boot.
3207 * The CPUAffinity= setting in service unit files now supports a new
3208 special value "numa" that causes the CPU affinity masked to be set
3209 based on the NUMA mask.
3211 * systemd will now log about all left-over processes remaining in a
3212 unit when the unit is stopped. It will now warn about services using
3213 KillMode=none, as this is generally an unsafe thing to make use of.
3215 * Two new unit file settings
3216 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
3217 added. They may be used to check whether a specific file system path
3218 resides on a block device that is encrypted on the block level
3219 (i.e. using dm-crypt/LUKS).
3221 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
3222 has been added that may be used for simple environment checks. This
3223 is particularly useful when passing in environment variables from a
3224 container manager (or from PAM in case of the systemd --user
3227 * .service unit files now accept a new setting CoredumpFilter= which
3228 allows configuration of the memory sections coredumps of the
3229 service's processes shall include.
3231 * .mount units gained a new ReadWriteOnly= boolean option. If set
3232 it will not be attempted to mount a file system read-only if mounting
3233 in read-write mode doesn't succeed. An option x-systemd.rw-only is
3234 available in /etc/fstab to control the same.
3236 * .socket units gained a new boolean setting PassPacketInfo=. If
3237 enabled, the kernel will attach additional per-packet metadata to all
3238 packets read from the socket, as an ancillary message. This controls
3239 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
3240 depending on socket type.
3242 * .service units gained a new setting RootHash= which may be used to
3243 specify the root hash for verity enabled disk images which are
3244 specified in RootImage=. RootVerity= may be used to specify a path to
3245 the Verity data matching a RootImage= file system. (The latter is
3246 only useful for images that do not contain the Verity data embedded
3247 into the same image that carries a GPT partition table following the
3248 Discoverable Partition Specification). Similarly, systemd-nspawn
3249 gained a new switch --verity-data= that takes a path to a file with
3250 the verity data of the disk image supplied in --image=, if the image
3251 doesn't contain the verity data itself.
3253 * .service units gained a new setting RootHashSignature= which takes
3254 either a base64 encoded PKCS#7 signature of the root hash specified
3255 with RootHash=, or a path to a file to read the signature from. This
3256 allows validation of the root hash against public keys available in
3257 the kernel keyring, and is only supported on recent kernels
3258 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
3259 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
3260 this mechanism has also been added to systemd-veritysetup.
3262 * .service unit files gained two new options
3263 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
3264 tune behaviour if a start or stop timeout is hit, i.e. whether to
3265 terminate the service with SIGTERM, SIGABRT or SIGKILL.
3267 * Most options in systemd that accept hexadecimal values prefixed with
3268 0x in additional to the usual decimal notation now also support octal
3269 notation when the 0o prefix is used and binary notation if the 0b
3272 * Various command line parameters and configuration file settings that
3273 configure key or certificate files now optionally take paths to
3274 AF_UNIX sockets in the file system. If configured that way a stream
3275 connection is made to the socket and the required data read from
3276 it. This is a simple and natural extension to the existing regular
3277 file logic, and permits other software to provide keys or
3278 certificates via simple IPC services, for example when unencrypted
3279 storage on disk is not desired. Specifically, systemd-networkd's
3280 Wireguard and MACSEC key file settings as well as
3281 systemd-journal-gatewayd's and systemd-journal-remote's PEM
3282 key/certificate parameters support this now.
3284 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
3285 configuration files that support specifier expansion learnt six new
3286 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
3287 resolve to the various ID fields from /etc/os-release, %l resolves to
3288 the "short" hostname of the system, i.e. the hostname configured in
3289 the kernel truncated at the first dot.
3291 * Support for the .include syntax in unit files has been removed. The
3292 concept has been obsolete for 6 years and we started warning about
3293 its pending removal 2 years ago (also see NEWS file below). It's
3296 * StandardError= and StandardOutput= in unit files no longer support
3297 the "syslog" and "syslog-console" switches. They were long removed
3298 from the documentation, but will now result in warnings when used,
3299 and be converted to "journal" and "journal+console" automatically.
3301 * If the service setting User= is set to the "nobody" user, a warning
3302 message is now written to the logs (but the value is nonetheless
3303 accepted). Setting User=nobody is unsafe, since the primary purpose
3304 of the "nobody" user is to own all files whose owner cannot be mapped
3305 locally. It's in particular used by the NFS subsystem and in user
3306 namespacing. By running a service under this user's UID it might get
3307 read and even write access to all these otherwise unmappable files,
3308 which is quite likely a major security problem.
3310 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
3311 and others) now have a size and inode limits applied (50% of RAM for
3312 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
3313 that the implicit kernel default is 50% too, so there is no change
3314 in the size limit for /tmp and /dev/shm.
3316 * nss-mymachines lost support for resolution of users and groups, and
3317 now only does resolution of hostnames. This functionality is now
3318 provided by nss-systemd. Thus, the 'mymachines' entry should be
3319 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
3320 (and 'systemd' added if it is not already there).
3322 * A new kernel command line option systemd.hostname= has been added
3323 that allows controlling the hostname that is initialized early during
3326 * A kernel command line option "udev.blockdev_read_only" has been
3327 added. If specified all hardware block devices that show up are
3328 immediately marked as read-only by udev. This option is useful for
3329 making sure that a specific boot under no circumstances modifies data
3330 on disk. Use "blockdev --setrw" to undo the effect of this, per
3333 * A new boolean kernel command line option systemd.swap= has been
3334 added, which may be used to turn off automatic activation of swap
3335 devices listed in /etc/fstab.
3337 * New kernel command line options systemd.condition-needs-update= and
3338 systemd.condition-first-boot= have been added, which override the
3339 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
3342 * A new kernel command line option systemd.clock-usec= has been added
3343 that allows setting the system clock to the specified time in µs
3344 since Jan 1st, 1970 early during boot. This is in particular useful
3345 in order to make test cases more reliable.
3347 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
3348 systemd-coredump to save core files for suid processes. When saving
3349 the core file, systemd-coredump will use the effective uid and gid of
3350 the process that faulted.
3352 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
3353 now automatically set to "Y" at boot, in order to enable pstore
3354 generation for collection with systemd-pstore.
3356 * We provide a set of udev rules to enable auto-suspend on PCI and USB
3357 devices that were tested to correctly support it. Previously, this
3358 was distributed as a set of udev rules, but has now been replaced by
3359 by a set of hwdb entries (and a much shorter udev rule to take action
3360 if the device modalias matches one of the new hwdb entries).
3362 As before, entries are periodically imported from the database
3363 maintained by the ChromiumOS project. If you have a device that
3364 supports auto-suspend correctly and where it should be enabled by
3365 default, please submit a patch that adds it to the database (see
3366 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
3368 * systemd-udevd gained the new configuration option timeout_signal= as well
3369 as a corresponding kernel command line option udev.timeout_signal=.
3370 The option can be used to configure the UNIX signal that the main
3371 daemon sends to the worker processes on timeout. Setting the signal
3372 to SIGABRT is useful for debugging.
3374 * .link files managed by systemd-udevd gained options RxFlowControl=,
3375 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
3376 order to configure various flow control parameters. They also gained
3377 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
3378 frame ring buffer sizes.
3380 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
3381 enabled systemd-networkd manages all routes configured by other tools.
3383 * .network files managed by systemd-networkd gained a new section
3384 [SR-IOV], in order to configure SR-IOV capable network devices.
3386 * systemd-networkd's [IPv6Prefix] section in .network files gained a
3387 new boolean setting Assign=. If enabled an address from the prefix is
3388 automatically assigned to the interface.
3390 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
3391 controls delegated prefixes assigned by DHCPv6 client. The section
3392 has three settings: SubnetID=, Assign=, and Token=. The setting
3393 SubnetID= allows explicit configuration of the preferred subnet that
3394 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
3395 Assign= is enabled (which is the default) an address from any acquired
3396 delegated prefix is automatically chosen and assigned to the
3397 interface. The setting Token= specifies an optional address generation
3400 * systemd-networkd's [Network] section gained a new setting
3401 IPv4AcceptLocal=. If enabled the interface accepts packets with local
3404 * systemd-networkd gained support for configuring the HTB queuing
3405 discipline in the [HierarchyTokenBucket] and
3406 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
3407 be configured in the [PFIFO] section, "GRED" in
3408 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
3409 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
3410 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
3411 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
3412 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
3413 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
3415 * systemd-networkd gained support for a new Termination= setting in the
3416 [CAN] section for configuring the termination resistor. It also
3417 gained a new ListenOnly= setting for controlling whether to only
3418 listen on CAN interfaces, without interfering with traffic otherwise
3419 (which is useful for debugging/monitoring CAN network
3420 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
3421 been added to configure various CAN-FD aspects.
3423 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
3424 When enabled, DHCPv6 will be attempted right-away without requiring an
3425 Router Advertisement packet suggesting it first (i.e. without the 'M'
3426 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
3427 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
3428 the RA packets suggest it.
3430 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
3431 which may be used to turn off use of the gateway information provided
3432 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
3433 used to configure how to process leases that lack a lifetime option.
3435 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
3436 setting SendVendorOption= allowing configuration of additional vendor
3437 options to send in the DHCP requests/responses. The [DHCPv6] section
3438 gained a new SendOption= setting for sending arbitrary DHCP
3439 options. RequestOptions= has been added to request arbitrary options
3440 from the server. UserClass= has been added to set the DHCP user class
3443 * systemd-networkd's [DHCPServer] section gained a new set of options
3444 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
3445 information about these three protocols in the DHCP lease. It also
3446 gained support for including "MUD" URLs ("Manufacturer Usage
3447 Description"). Support for "MUD" URLs was also added to the LLDP
3448 stack, configurable in the [LLDP] section in .network files.
3450 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
3451 mode. Also, the sections now support a new setting SourceMACAddress=.
3453 * systemd-networkd's .netdev files now support a new setting
3454 VLANProtocol= in the [Bridge] section that allows configuration of
3455 the VLAN protocol to use.
3457 * systemd-networkd supports a new Group= setting in the [Link] section
3458 of the .network files, to control the link group.
3460 * systemd-networkd's [Network] section gained a new
3461 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
3462 link local address is generated.
3464 * A new default .network file is now shipped that matches TUN/TAP
3465 devices that begin with "vt-" in their name. Such interfaces will
3466 have IP routing onto the host links set up automatically. This is
3467 supposed to be used by VM managers to trivially acquire a network
3468 interface which is fully set up for host communication, simply by
3469 carefully picking an interface name to use.
3471 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
3472 which sets the route priority for routes specified by the DHCP server.
3474 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
3475 which configures the vendor class information sent to DHCP server.
3477 * The BlackList= settings in .network files' [DHCPv4] and
3478 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
3479 are still understood to provide compatibility.
3481 * networkctl gained the new "forcerenew" command for forcing all DHCP
3482 server clients to renew their lease. The interface "status" output
3483 will now show numerous additional fields of information about an
3484 interface. There are new "up" and "down" commands to bring specific
3485 interfaces up or down.
3487 * systemd-resolved's DNS= configuration option now optionally accepts a
3488 port number (after ":") and a host name (after "#"). When the host
3489 name is specified, the DNS-over-TLS certificate is validated to match
3490 the specified hostname. Additionally, in case of IPv6 addresses, an
3491 interface may be specified (after "%").
3493 * systemd-resolved may be configured to forward single-label DNS names.
3494 This is not standard-conformant, but may make sense in setups where
3495 public DNS servers are not used.
3497 * systemd-resolved's DNS-over-TLS support gained SNI validation.
3499 * systemd-nspawn's --resolv-conf= switch gained a number of new
3500 supported values. Specifically, options starting with "replace-" are
3501 like those prefixed "copy-" but replace any existing resolv.conf
3502 file. And options ending in "-uplink" and "-stub" can now be used to
3503 propagate other flavours of resolv.conf into the container (as
3504 defined by systemd-resolved).
3506 * The various programs included in systemd can now optionally output
3507 their log messages on stderr prefixed with a timestamp, controlled by
3508 the $SYSTEMD_LOG_TIME environment variable.
3510 * systemctl gained a new "-P" switch that is a shortcut for "--value
3513 * "systemctl list-units" and "systemctl list-machines" no longer hide
3514 their first output column with --no-legend. To hide the first column,
3517 * "systemctl reboot" takes the option "--reboot-argument=".
3518 The optional positional argument to "systemctl reboot" is now
3519 being deprecated in favor of this option.
3521 * systemd-run gained a new switch --slice-inherit. If specified the
3522 unit it generates is placed in the same slice as the systemd-run
3525 * systemd-journald gained support for zstd compression of large fields
3526 in journal files. The hash tables in journal files have been hardened
3527 against hash collisions. This is an incompatible change and means
3528 that journal files created with new systemd versions are not readable
3529 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
3530 environment variable for systemd-journald.service is set to 0 this
3531 new hardening functionality may be turned off, so that generated
3532 journal files remain compatible with older journalctl
3535 * journalctl will now include a clickable link in the default output for
3536 each log message for which an URL with further documentation is
3537 known. This is only supported on terminal emulators that support
3538 clickable hyperlinks, and is turned off if a pager is used (since
3539 "less" still doesn't support hyperlinks,
3540 unfortunately). Documentation URLs may be included in log messages
3541 either by including a DOCUMENTATION= journal field in it, or by
3542 associating a journal message catalog entry with the log message's
3543 MESSAGE_ID, which then carries a "Documentation:" tag.
3545 * journald.conf gained a new boolean setting Audit= that may be used to
3546 control whether systemd-journald will enable audit during
3549 * when systemd-journald's log stream is broken up into multiple lines
3550 because the PID of the sender changed this is indicated in the
3551 generated log records via the _LINE_BREAK=pid-change field.
3553 * journalctl's "-o cat" output mode will now show one or more journal
3554 fields specified with --output-fields= instead of unconditionally
3555 MESSAGE=. This is useful to retrieve a very specific set of fields
3556 without any decoration.
3558 * The sd-journal.h API gained two new functions:
3559 sd_journal_enumerate_available_unique() and
3560 sd_journal_enumerate_available_data() that operate like their
3561 counterparts that lack the _available_ in the name, but skip items
3562 that cannot be read and processed by the local implementation
3563 (i.e. are compressed in an unsupported format or such),
3565 * coredumpctl gained a new --file= switch, matching the same one in
3566 journalctl: a specific journal file may be specified to read the
3569 * coredumps collected by systemd-coredump may now be compressed using
3572 * systemd-binfmt gained a new switch --unregister for unregistering all
3573 registered entries at once. This is now invoked automatically at
3574 shutdown, so that binary formats registered with the "F" flag will
3575 not block clean file system unmounting.
3577 * systemd-notify's --pid= switch gained new values: "parent", "self",
3578 "auto" for controlling which PID to send to the service manager: the
3579 systemd-notify process' PID, or the one of the process invoking it.
3581 * systemd-logind's Session bus object learnt a new method call
3582 SetType() for temporarily updating the session type of an already
3583 allocated session. This is useful for upgrading tty sessions to
3584 graphical ones once a compositor is invoked.
3586 * systemd-socket-proxy gained a new switch --exit-idle-time= for
3587 configuring an exit-on-idle time.
3589 * systemd-repart's --empty= setting gained a new value "create". If
3590 specified a new empty regular disk image file is created under the
3591 specified name. Its size may be specified with the new --size=
3592 option. The latter is also supported without the "create" mode, in
3593 order to grow existing disk image files to the specified size. These
3594 two new options are useful when creating or manipulating disk images
3595 instead of operating on actual block devices.
3597 * systemd-repart drop-ins now support a new UUID= setting to control
3598 the UUID to assign to a newly created partition.
3600 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
3603 * systemd-repart's Label= setting now support the usual, simple
3604 specifier expansion.
3606 * systemd-homed's LUKS backend gained the ability to discard empty file
3607 system blocks automatically when the user logs out. This is enabled
3608 by default to ensure that home directories take minimal space when
3609 logged out but get full size guarantees when logged in. This may be
3610 controlled with the new --luks-offline-discard= switch to homectl.
3612 * If systemd-homed detects that /home/ is encrypted as a whole it will
3613 now default to the directory or subvolume backends instead of the
3614 LUKS backend, in order to avoid double encryption. The default
3615 storage and file system may now be configured explicitly, too, via
3616 the new /etc/systemd/homed.conf configuration file.
3618 * systemd-homed now supports unlocking home directories with FIDO2
3619 security tokens that support the 'hmac-secret' extension, in addition
3620 to the existing support for PKCS#11 security token unlocking
3621 support. Note that many recent hardware security tokens support both
3622 interfaces. The FIDO2 support is accessible via homectl's
3623 --fido2-device= option.
3625 * homectl's --pkcs11-uri= setting now accepts two special parameters:
3626 if "auto" is specified and only one suitable PKCS#11 security token
3627 is plugged in, its URL is automatically determined and enrolled for
3628 unlocking the home directory. If "list" is specified a brief table of
3629 suitable PKCS#11 security tokens is shown. Similar, the new
3630 --fido2-device= option also supports these two special values, for
3631 automatically selecting and listing suitable FIDO2 devices.
3633 * The /etc/crypttab tmp option now optionally takes an argument
3634 selecting the file system to use. Moreover, the default is now
3635 changed from ext2 to ext4.
3637 * There's a new /etc/crypttab option "keyfile-erase". If specified the
3638 key file listed in the same line is removed after use, regardless if
3639 volume activation was successful or not. This is useful if the key
3640 file is only acquired transiently at runtime and shall be erased
3641 before the system continues to boot.
3643 * There's also a new /etc/crypttab option "try-empty-password". If
3644 specified, before asking the user for a password it is attempted to
3645 unlock the volume with an empty password. This is useful for
3646 installing encrypted images whose password shall be set on first boot
3647 instead of at installation time.
3649 * systemd-cryptsetup will now attempt to load the keys to unlock
3650 volumes with automatically from files in
3651 /etc/cryptsetup-keys.d/<volume>.key and
3652 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
3654 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
3655 /etc/crypttab, during boot.
3657 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
3658 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
3661 * A new generator systemd-xdg-autostart-generator has been added. It
3662 generates systemd unit files from XDG autostart .desktop files, and
3663 may be used to let the systemd user instance manage services that are
3664 started automatically as part of the desktop session.
3666 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
3667 to query and change the firmware's 'reboot into firmware' setup flag.
3669 * systemd-firstboot gained a new switch --kernel-command-line= that may
3670 be used to initialize the /etc/kernel/cmdline file of the image. It
3671 also gained a new switch --root-password-hashed= which is like
3672 --root-password= but accepts a pre-hashed UNIX password as
3673 argument. The new option --delete-root-password may be used to unset
3674 any password for the root user (dangerous!). The --root-shell= switch
3675 may be used to control the shell to use for the root account. A new
3676 --force option may be used to override any already set settings with
3677 the parameters specified on the command line (by default, the tool
3678 will not override what has already been set before, i.e. is purely
3681 * systemd-firstboot gained support for a new --image= switch, which is
3682 similar to --root= but accepts the path to a disk image file, on
3683 which it then operates.
3685 * A new sd-path.h API has been added to libsystemd. It provides a
3686 simple API for retrieving various search paths and primary
3687 directories for various resources.
3689 * A new call sd_notify_barrier() has been added to the sd-daemon.h
3690 API. The call will block until all previously sent sd_notify()
3691 messages have been processed by the service manager. This is useful
3692 to remove races caused by a process already having disappeared at the
3693 time a notification message is processed by the service manager,
3694 making correct attribution impossible. The systemd-notify tool will
3695 now make use of this call implicitly, but this can be turned off again
3696 via the new --no-block switch.
3698 * When sending a file descriptor (fd) to the service manager to keep
3699 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
3700 may be specified. If passed the service manager will refrain from
3701 poll()ing on the file descriptor. Traditionally (and when the
3702 parameter is not specified), the service manager will poll it for
3703 POLLHUP or POLLERR events, and immediately close the fds in that
3706 * The service manager (PID1) gained a new D-Bus method call
3707 SetShowStatus() which may be used to control whether it shall show
3708 boot-time status output on the console. This method has a similar
3709 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
3711 * The sd-bus API gained a number of convenience functions that take
3712 va_list arguments rather than "...". For example, there's now
3713 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
3714 it easier to build wrappers that accept variadic arguments and want
3715 to pass a ready va_list structure to sd-bus.
3717 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
3718 flag which alters how the userdata pointer to pass to the callbacks
3719 is determined. When the flag is set, the offset field is converted
3720 as-is into a pointer, without adding it to the object pointer the
3721 vtable is associated with.
3723 * sd-bus now exposes four new functions:
3724 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
3725 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
3726 validate strings to check if they qualify as various D-Bus concepts.
3728 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
3729 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
3730 that simplify adding argument names to D-Bus methods and signals.
3732 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
3734 * Various D-Bus APIs of systemd daemons now have man pages that
3735 document the methods, signals and properties.
3737 * The expectations on user/group name syntax are now documented in
3738 detail; documentation on how classic home directories may be
3739 converted into home directories managed by homed has been added;
3740 documentation regarding integration of homed/userdb functionality in
3741 desktops has been added:
3743 https://systemd.io/USER_NAMES
3744 https://systemd.io/CONVERTING_TO_HOMED
3745 https://systemd.io/USERDB_AND_DESKTOPS
3747 * Documentation for the on-disk Journal file format has been updated
3748 and has now moved to:
3750 https://systemd.io/JOURNAL_FILE_FORMAT
3752 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
3753 has been extended by a set of environment variables that expose
3754 select fields from the host's os-release file to the container
3755 payload. Similarly, host's os-release files can be mounted into the
3756 container underneath /run/host. Together, those mechanisms provide a
3757 standardized way to expose information about the host to the
3758 container payload. Both interfaces are implemented in systemd-nspawn.
3760 * All D-Bus services shipped in systemd now implement the generic
3761 LogControl1 D-Bus API which allows clients to change log level +
3762 target of the service during runtime.
3764 * Only relevant for developers: the mkosi.default symlink has been
3765 dropped from version control. Please create a symlink to one of the
3766 distribution-specific defaults in .mkosi/ based on your preference.
3768 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
3769 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
3770 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
3771 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
3772 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
3773 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
3774 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
3775 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
3776 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
3777 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
3778 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
3779 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
3780 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
3781 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
3782 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
3783 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
3784 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
3785 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
3786 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
3787 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
3788 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
3789 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
3790 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
3791 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
3792 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
3793 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
3794 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
3795 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
3796 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
3797 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
3798 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
3799 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
3800 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
3801 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
3802 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
3803 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
3804 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
3805 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
3807 – Warsaw, 2020-07-30
3811 * A new tool "systemd-repart" has been added, that operates as an
3812 idempotent declarative repartitioner for GPT partition tables.
3813 Specifically, a set of partitions that must or may exist can be
3814 configured via drop-in files, and during every boot the partition
3815 table on disk is compared with these files, creating missing
3816 partitions or growing existing ones based on configurable relative
3817 and absolute size constraints. The tool is strictly incremental,
3818 i.e. does not delete, shrink or move partitions, but only adds and
3819 grows them. The primary use-case is OS images that ship in minimized
3820 form, that on first boot are grown to the size of the underlying
3821 block device or augmented with additional partitions. For example,
3822 the root partition could be extended to cover the whole disk, or a
3823 swap or /home partitions could be added on first boot. It can also be
3824 used for systems that use an A/B update scheme but ship images with
3825 just the A partition, with B added on first boot. The tool is
3826 primarily intended to be run in the initrd, shortly before
3827 transitioning into the host OS, but can also be run after the
3828 transition took place. It automatically discovers the disk backing
3829 the root file system, and should hence not require any additional
3830 configuration besides the partition definition drop-ins. If no
3831 configuration drop-ins are present, no action is taken.
3833 * A new component "userdb" has been added, along with a small daemon
3834 "systemd-userdbd.service" and a client tool "userdbctl". The framework
3835 allows defining rich user and group records in a JSON format,
3836 extending on the classic "struct passwd" and "struct group"
3837 structures. Various components in systemd have been updated to
3838 process records in this format, including systemd-logind and
3839 pam-systemd. The user records are intended to be extensible, and
3840 allow setting various resource management, security and runtime
3841 parameters that shall be applied to processes and sessions of the
3842 user as they log in. This facility is intended to allow associating
3843 such metadata directly with user/group records so that they can be
3844 produced, extended and consumed in unified form. We hope that
3845 eventually frameworks such as sssd will generate records this way, so
3846 that for the first time resource management and various other
3847 per-user settings can be configured in LDAP directories and then
3848 provided to systemd (specifically to systemd-logind and pam-system)
3849 to apply on login. For further details see:
3851 https://systemd.io/USER_RECORD
3852 https://systemd.io/GROUP_RECORD
3853 https://systemd.io/USER_GROUP_API
3855 * A small new service systemd-homed.service has been added, that may be
3856 used to securely manage home directories with built-in encryption.
3857 The complete user record data is unified with the home directory,
3858 thus making home directories naturally migratable. Its primary
3859 back-end is based on LUKS volumes, but fscrypt, plain directories,
3860 and other storage schemes are also supported. This solves a couple of
3861 problems we saw with traditional ways to manage home directories, in
3862 particular when it comes to encryption. For further discussion of
3863 this, see the video of Lennart's talk at AllSystemsGo! 2019:
3865 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
3867 For further details about the format and expectations on home
3868 directories this new daemon makes, see:
3870 https://systemd.io/HOME_DIRECTORY
3872 * systemd-journald is now multi-instantiable. In addition to the main
3873 instance systemd-journald.service there's now a template unit
3874 systemd-journald@.service, with each instance defining a new named
3875 log 'namespace' (whose name is specified via the instance part of the
3876 unit name). A new unit file setting LogNamespace= has been added,
3877 taking such a namespace name, that assigns services to the specified
3878 log namespaces. As each log namespace is serviced by its own
3879 independent journal daemon, this functionality may be used to improve
3880 performance and increase isolation of applications, at the price of
3881 losing global message ordering. Each instance of journald has a
3882 separate set of configuration files, with possibly different disk
3883 usage limitations and other settings.
3885 journalctl now takes a new option --namespace= to show logs from a
3886 specific log namespace. The sd-journal.h API gained
3887 sd_journal_open_namespace() for opening the log stream of a specific
3888 log namespace. systemd-journald also gained the ability to exit on
3889 idle, which is useful in the context of log namespaces, as this means
3890 log daemons for log namespaces can be activated automatically on
3891 demand and will stop automatically when no longer used, minimizing
3894 * When systemd-tmpfiles copies a file tree using the 'C' line type it
3895 will now label every copied file according to the SELinux database.
3897 * When systemd/PID 1 detects it is used in the initrd it will now boot
3898 into initrd.target rather than default.target by default. This should
3899 make it simpler to build initrds with systemd as for many cases the
3900 only difference between a host OS image and an initrd image now is
3901 the presence of the /etc/initrd-release file.
3903 * A new kernel command line option systemd.cpu_affinity= is now
3904 understood. It's equivalent to the CPUAffinity= option in
3905 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
3906 itself and the default for all other processes.
3908 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
3909 equivalent), the SELinux database is now reloaded, ensuring that
3910 sockets and other file system objects are generated taking the new
3911 database into account.
3913 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
3914 "quiet" has been changed to imply that instead of
3915 "systemd.show-status=auto". In this mode, only messages about errors
3916 and significant delays in boot are shown on the console.
3918 * The sd-event.h API gained native support for the new Linux "pidfd"
3919 concept. This permits watching processes using file descriptors
3920 instead of PID numbers, which fixes a number of races and makes
3921 process supervision more robust and efficient. All of systemd's
3922 components will now use pidfds if the kernel supports it for process
3923 watching, with the exception of PID 1 itself, unfortunately. We hope
3924 to move PID 1 to exclusively using pidfds too eventually, but this
3925 requires some more kernel work first. (Background: PID 1 watches
3926 processes using waitid() with the P_ALL flag, and that does not play
3927 together nicely with pidfds yet.)
3929 * Closely related to this, the sd-event.h API gained two new calls
3930 sd_event_source_send_child_signal() (for sending a signal to a
3931 watched process) and sd_event_source_get_child_process_own() (for
3932 marking a process so that it is killed automatically whenever the
3933 event source watching it is freed).
3935 * systemd-networkd gained support for configuring Token Bucket Filter
3936 (TBF) parameters in its qdisc configuration support. Similarly,
3937 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
3938 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
3940 * systemd-networkd gained support for Intermediate Functional Block
3941 (IFB) network devices.
3943 * systemd-networkd gained support for configuring multi-path IP routes,
3944 using the new MultiPathRoute= setting in the [Route] section.
3946 * systemd-networkd's DHCPv4 client has been updated to support a new
3947 SendDecline= option. If enabled, duplicate address detection is done
3948 after a DHCP offer is received from the server. If a conflict is
3949 detected, the address is declined. The DHCPv4 client also gained
3950 support for a new RouteMTUBytes= setting that allows to configure the
3951 MTU size to be used for routes generated from DHCPv4 leases.
3953 * The PrefixRoute= setting in systemd-networkd's [Address] section of
3954 .network files has been deprecated, and replaced by AddPrefixRoute=,
3955 with its sense inverted.
3957 * The Gateway= setting of [Route] sections of .network files gained
3958 support for a special new value "_dhcp". If set, the configured
3959 static route uses the gateway host configured via DHCP.
3961 * New User= and SuppressPrefixLength= settings have been implemented
3962 for the [RoutingPolicyRule] section of .network files to configure
3963 source routing based on UID ranges and prefix length, respectively.
3965 * The Type= match property of .link files has been generalized to
3966 always match the device type shown by 'networkctl status', even for
3967 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
3970 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
3971 D-Bus message object as "sensitive". Those objects are erased from
3972 memory when they are freed. This concept is intended to be used for
3973 messages that contain security sensitive data. A new flag
3974 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
3975 in sd-bus vtables, causing any incoming and outgoing messages of
3976 those methods to be implicitly marked as "sensitive".
3978 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
3979 contents of a message (or parts thereof) to standard output for
3982 * systemd-sysusers gained support for creating users with the primary
3983 group named differently than the user.
3985 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
3986 gained support for growing XFS partitions. Previously it supported
3987 only ext4 and btrfs partitions.
3989 * The support for /etc/crypttab gained a new x-initrd.attach option. If
3990 set, the specified encrypted volume is unlocked already in the
3991 initrd. This concept corresponds to the x-initrd.mount option in
3994 * systemd-cryptsetup gained native support for unlocking encrypted
3995 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
3996 encryption of volumes to YubiKeys. This is exposed in the new
3997 pkcs11-uri= option in /etc/crypttab.
3999 * The /etc/fstab support in systemd now supports two new mount options
4000 x-systemd.{required,wanted}-by=, for explicitly configuring the units
4001 that the specified mount shall be pulled in by, in place of
4002 the usual local-fs.target/remote-fs.target.
4004 * The https://systemd.io/ web site has been relaunched, directly
4005 populated with most of the documentation included in the systemd
4006 repository. systemd also acquired a new logo, thanks to Tobias
4009 * systemd-udevd gained support for managing "alternative" network
4010 interface names, as supported by new Linux kernels. For the first
4011 time this permits assigning multiple (and longer!) names to a network
4012 interface. systemd-udevd will now by default assign the names
4013 generated via all supported naming schemes to each interface. This
4014 may be further tweaked with .link files and the AlternativeName= and
4015 AlternativeNamesPolicy= settings. Other components of systemd have
4016 been updated to support the new alternative names wherever
4017 appropriate. For example, systemd-nspawn will now generate
4018 alternative interface names for the host-facing side of container
4019 veth links based on the full container name without truncation.
4021 * systemd-nspawn interface naming logic has been updated in another way
4022 too: if the main interface name (i.e. as opposed to new-style
4023 "alternative" names) based on the container name is truncated, a
4024 simple hashing scheme is used to give different interface names to
4025 multiple containers whose names all begin with the same prefix. Since
4026 this changes the primary interface names pointing to containers if
4027 truncation happens, the old scheme may still be requested by
4028 selecting an older naming scheme, via the net.naming-scheme= kernel
4029 command line option.
4031 * PrivateUsers= in service files now works in services run by the
4032 systemd --user per-user instance of the service manager.
4034 * A new per-service sandboxing option ProtectClock= has been added that
4035 locks down write access to the system clock. It takes away device
4036 node access to /dev/rtc as well as the system calls that set the
4037 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
4038 Note that this option does not affect access to auxiliary services
4039 that allow changing the clock, for example access to
4042 * The systemd-id128 tool gained a new "show" verb for listing or
4043 resolving a number of well-known UUIDs/128bit IDs, currently mostly
4044 GPT partition table types.
4046 * The Discoverable Partitions Specification has been updated to support
4047 /var and /var/tmp partition discovery. Support for this has been
4048 added to systemd-gpt-auto-generator. For details see:
4050 https://systemd.io/DISCOVERABLE_PARTITIONS
4052 * "systemctl list-unit-files" has been updated to show a new column
4053 with the suggested enablement state based on the vendor preset files
4054 for the respective units.
4056 * "systemctl" gained a new option "--with-dependencies". If specified
4057 commands such as "systemctl status" or "systemctl cat" will now show
4058 all specified units along with all units they depend on.
4060 * networkctl gained support for showing per-interface logs in its
4063 * systemd-networkd-wait-online gained support for specifying the maximum
4064 operational state to wait for, and to wait for interfaces to
4067 * The [Match] section of .link and .network files now supports a new
4068 option PermanentMACAddress= which may be used to check against the
4069 permanent MAC address of a network device even if a randomized MAC
4072 * The [TrafficControlQueueingDiscipline] section in .network files has
4073 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
4074 dropped from the individual setting names.
4076 * Any .link and .network files that have an empty [Match] section (this
4077 also includes empty and commented-out files) will now be
4078 rejected. systemd-udev and systemd-networkd started warning about
4079 such files in version 243.
4081 * systemd-logind will now validate access to the operation of changing
4082 the virtual terminal via a polkit action. By default, only users
4083 with at least one session on a local VT are granted permission.
4085 * When systemd sets up PAM sessions that invoked service processes
4086 shall run in, the pam_setcred() API is now invoked, thus permitting
4087 PAM modules to set additional credentials for the processes.
4089 * portablectl attach/detach verbs now accept --now and --enable options
4090 to combine attachment with enablement and invocation, or detachment
4091 with stopping and disablement.
4093 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
4094 fixed, which in turn exposed bugs in unit configuration of services
4095 which have Type=oneshot and should only run once, but do not have
4096 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
4097 service may be started again after exiting successfully, for example
4098 as a dependency in another transaction. Affected services included
4099 some internal systemd services (most notably
4100 systemd-vconsole-setup.service, which was updated to have
4101 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
4102 plymouth has been suitably updated or patched before upgrading to
4103 this systemd release. See
4104 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
4105 additional discussion.
4107 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
4108 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
4109 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
4110 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
4111 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
4112 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
4113 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
4114 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
4115 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
4116 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
4117 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
4118 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
4119 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
4120 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
4121 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
4122 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
4123 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
4124 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
4125 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
4126 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
4127 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
4128 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
4129 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
4130 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
4133 – Warsaw, 2020-03-06
4137 * Support for the cpuset cgroups v2 controller has been added.
4138 Processes may be restricted to specific CPUs using the new
4139 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
4140 AllowedMemoryNodes= setting.
4142 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
4143 now be configured using a new RestartKillSignal= setting. This
4144 allows units which signals to request termination to implement
4145 different behaviour when stopping in preparation for a restart.
4147 * "systemctl clean" may now be used also for socket, mount, and swap
4150 * systemd will also read configuration options from the EFI variable
4151 SystemdOptions. This may be used to configure systemd behaviour when
4152 modifying the kernel command line is inconvenient, but configuration
4153 on disk is read too late, for example for the options related to
4154 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
4155 set the EFI variable.
4157 * systemd will now disable printk ratelimits in early boot. This should
4158 allow us to capture more logs from the early boot phase where normal
4159 storage is not available and the kernel ring buffer is used for
4160 logging. Configuration on the kernel command line has higher priority
4161 and overrides the systemd setting.
4163 systemd programs which log to /dev/kmsg directly use internal
4164 ratelimits to prevent runaway logging. (Normally this is only used
4165 during early boot, so in practice this change has very little
4168 * Unit files now support top level dropin directories of the form
4169 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
4170 that affects all corresponding unit files.
4172 * systemctl gained support for 'stop --job-mode=triggering' which will
4173 stop the specified unit and any units which could trigger it.
4175 * Unit status display now includes units triggering and triggered by
4176 the unit being shown.
4178 * The RuntimeMaxSec= setting is now supported by scopes, not just
4179 .service units. This is particularly useful for PAM sessions which
4180 create a scope unit for the user login. systemd.runtime_max_sec=
4181 setting may used with the pam_systemd module to limit the duration
4182 of the PAM session, for example for time-limited logins.
4184 * A new @pkey system call group is now defined to make it easier to
4185 allow-list memory protection syscalls for containers and services
4186 which need to use them.
4188 * systemd-udevd: removed the 30s timeout for killing stale workers on
4189 exit. systemd-udevd now waits for workers to finish. The hard-coded
4190 exit timeout of 30s was too short for some large installations, where
4191 driver initialization could be prematurely interrupted during initrd
4192 processing if the root file system had been mounted and init was
4193 preparing to switch root. If udevd is run without systemd and workers
4194 are hanging while udevd receives an exit signal, udevd will now exit
4195 when udev.event_timeout is reached for the last hanging worker. With
4196 systemd, the exit timeout can additionally be configured using
4197 TimeoutStopSec= in systemd-udevd.service.
4199 * udev now provides a program (fido_id) that identifies FIDO CTAP1
4200 ("U2F")/CTAP2 security tokens based on the usage declared in their
4201 report and descriptor and outputs suitable environment variables.
4202 This replaces the externally maintained allow lists of all known
4203 security tokens that were used previously.
4205 * Automatically generated autosuspend udev rules for allow-listed
4206 devices have been imported from the Chromium OS project. This should
4207 improve power saving with many more devices.
4209 * udev gained a new "CONST{key}=value" setting that allows matching
4210 against system-wide constants without forking a helper binary.
4211 Currently "arch" and "virt" keys are supported.
4213 * udev now opens CDROMs in non-exclusive mode when querying their
4214 capabilities. This should fix issues where other programs trying to
4215 use the CDROM cannot gain access to it, but carries a risk of
4216 interfering with programs writing to the disk, if they did not open
4217 the device in exclusive mode as they should.
4219 * systemd-networkd does not create a default route for IPv4 link local
4220 addressing anymore. The creation of the route was unexpected and was
4221 breaking routing in various cases, but people who rely on it being
4222 created implicitly will need to adjust. Such a route may be requested
4223 with DefaultRouteOnDevice=yes.
4225 Similarly, systemd-networkd will not assign a link-local IPv6 address
4226 when IPv6 link-local routing is not enabled.
4228 * Receive and transmit buffers may now be configured on links with
4229 the new RxBufferSize= and TxBufferSize= settings.
4231 * systemd-networkd may now advertise additional IPv6 routes. A new
4232 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
4235 * systemd-networkd may now configure "next hop" routes using the
4236 [NextHop] section and Gateway= and Id= settings.
4238 * systemd-networkd will now retain DHCP config on restarts by default
4239 (but this may be overridden using the KeepConfiguration= setting).
4240 The default for SendRelease= has been changed to true.
4242 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
4243 received from the server.
4245 The client will use the received SIP server list if UseSIP=yes is
4248 The client may be configured to request specific options from the
4249 server using a new RequestOptions= setting.
4251 The client may be configured to send arbitrary options to the server
4252 using a new SendOption= setting.
4254 A new IPServiceType= setting has been added to configure the "IP
4255 service type" value used by the client.
4257 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
4258 request prefix hints in the DHCPv6 solicitation.
4260 * The DHCPv4 server may be configured to send arbitrary options using
4261 a new SendOption= setting.
4263 * The DHCPv4 server may now be configured to emit SIP server list using
4264 the new EmitSIP= and SIP= settings.
4266 * systemd-networkd and networkctl may now renew DHCP leases on demand.
4267 networkctl has a new 'networkctl renew' verb.
4269 * systemd-networkd may now reconfigure links on demand. networkctl
4270 gained two new verbs: "reload" will reload the configuration, and
4271 "reconfigure DEVICE…" will reconfigure one or more devices.
4273 * .network files may now match on SSID and BSSID of a wireless network,
4274 i.e. the access point name and hardware address using the new SSID=
4275 and BSSID= options. networkctl will display the current SSID and
4276 BSSID for wireless links.
4278 .network files may also match on the wireless network type using the
4279 new WLANInterfaceType= option.
4281 * systemd-networkd now includes default configuration that enables
4282 link-local addressing when connected to an ad-hoc wireless network.
4284 * systemd-networkd may configure the Traffic Control queueing
4285 disciplines in the kernel using the new
4286 [TrafficControlQueueingDiscipline] section and Parent=,
4287 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
4288 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
4289 NetworkEmulatorDuplicateRate= settings.
4291 * systemd-tmpfiles gained a new w+ setting to append to files.
4293 * systemd-analyze dump will now report when the memory configuration in
4294 the kernel does not match what systemd has configured (usually,
4295 because some external program has modified the kernel configuration
4298 * systemd-analyze gained a new --base-time= switch instructs the
4299 'calendar' verb to resolve times relative to that timestamp instead
4300 of the present time.
4302 * journalctl --update-catalog now produces deterministic output (making
4303 reproducible image builds easier).
4305 * A new devicetree-overlay setting is now documented in the Boot Loader
4308 * The default value of the WatchdogSec= setting used in systemd
4309 services (the ones bundled with the project itself) may be set at
4310 configuration time using the -Dservice-watchdog= setting. If set to
4311 empty, the watchdogs will be disabled.
4313 * systemd-resolved validates IP addresses in certificates now when GnuTLS
4316 * libcryptsetup >= 2.0.1 is now required.
4318 * A configuration option -Duser-path= may be used to override the $PATH
4319 used by the user service manager. The default is again to use the same
4320 path as the system manager.
4322 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
4323 outputting the 128bit IDs in UUID format (i.e. in the "canonical
4326 * Service units gained a new sandboxing option ProtectKernelLogs= which
4327 makes sure the program cannot get direct access to the kernel log
4328 buffer anymore, i.e. the syslog() system call (not to be confused
4329 with the API of the same name in libc, which is not affected), the
4330 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
4331 inaccessible to the service. It's recommended to enable this setting
4332 for all services that should not be able to read from or write to the
4333 kernel log buffer, which are probably almost all.
4335 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
4336 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
4337 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
4338 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
4339 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
4340 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
4341 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
4342 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
4343 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
4344 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
4345 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
4346 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
4347 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
4348 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
4349 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
4350 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
4351 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
4352 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
4353 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
4354 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
4355 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
4356 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
4357 Zach Smith, Zbigniew Jędrzejewski-Szmek
4359 – Warsaw, 2019-11-29
4363 * This release enables unprivileged programs (i.e. requiring neither
4364 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
4365 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
4366 kernel for the whole UNIX group range, i.e. all processes. This
4367 change should be reasonably safe, as the kernel support for it was
4368 specifically implemented to allow safe access to ICMP Echo for
4369 processes lacking any privileges. If this is not desirable, it can be
4370 disabled again by setting the parameter to "1 0".
4372 * Previously, filters defined with SystemCallFilter= would have the
4373 effect that any calling of an offending system call would terminate
4374 the calling thread. This behaviour never made much sense, since
4375 killing individual threads of unsuspecting processes is likely to
4376 create more problems than it solves. With this release the default
4377 action changed from killing the thread to killing the whole
4378 process. For this to work correctly both a kernel version (>= 4.14)
4379 and a libseccomp version (>= 2.4.0) supporting this new seccomp
4380 action is required. If an older kernel or libseccomp is used the old
4381 behaviour continues to be used. This change does not affect any
4382 services that have no system call filters defined, or that use
4383 SystemCallErrorNumber= (and thus see EPERM or another error instead
4384 of being killed when calling an offending system call). Note that
4385 systemd documentation always claimed that the whole process is
4386 killed. With this change behaviour is thus adjusted to match the
4389 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
4390 4194304 by default, i.e. the full 22bit range the kernel allows, up
4391 from the old 16bit range. This should improve security and
4392 robustness, as PID collisions are made less likely (though certainly
4393 still possible). There are rumours this might create compatibility
4394 problems, though at this moment no practical ones are known to
4395 us. Downstream distributions are hence advised to undo this change in
4396 their builds if they are concerned about maximum compatibility, but
4397 for everybody else we recommend leaving the value bumped. Besides
4398 improving security and robustness this should also simplify things as
4399 the maximum number of allowed concurrent tasks was previously bounded
4400 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
4401 only a single knob is left ("kernel.threads-max"). There have been
4402 concerns that usability is affected by this change because larger PID
4403 numbers are harder to type, but we believe the change from 5 digits
4404 to 7 digits doesn't hamper usability.
4406 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
4407 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
4408 hierarchically set default memory protection values for a particular
4409 subtree of the unit hierarchy.
4411 * Memory protection directives can now take a value of zero, allowing
4412 explicit opting out of a default value propagated by an ancestor.
4414 * systemd now defaults to the "unified" cgroup hierarchy setup during
4415 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
4416 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
4417 change reflects the fact that cgroupsv2 support has matured
4418 substantially in both systemd and in the kernel, and is clearly the
4419 way forward. Downstream production distributions might want to
4420 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
4421 their builds as unfortunately the popular container managers have not
4422 caught up with the kernel API changes.
4424 * Man pages are not built by default anymore (html pages were already
4425 disabled by default), to make development builds quicker. When
4426 building systemd for a full installation with documentation, meson
4427 should be called with -Dman=true and/or -Dhtml=true as appropriate.
4428 The default was changed based on the assumption that quick one-off or
4429 repeated development builds are much more common than full optimized
4430 builds for installation, and people need to pass various other
4431 options to when doing "proper" builds anyway, so the gain from making
4432 development builds quicker is bigger than the one time disruption for
4435 Two scripts are created in the *build* directory to generate and
4436 preview man and html pages on demand, e.g.:
4438 build/man/man systemctl
4439 build/man/html systemd.index
4441 * libidn2 is used by default if both libidn2 and libidn are installed.
4442 Please use -Dlibidn=true if libidn is preferred.
4444 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
4445 big-endian machines. Before, bytes were written and read in native
4446 machine order as exposed by the native libc __cpu_mask interface.
4447 Now, little-endian order is always used (CPUs 0–7 are described by
4448 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
4449 This change fixes D-Bus calls that cross endianness boundary.
4451 The presentation format used for CPUAffinity= by "systemctl show" and
4452 "systemd-analyze dump" is changed to present CPU indices instead of
4453 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
4454 shown as CPUAffinity=03000000000000000000000000000… (on
4455 little-endian) or CPUAffinity=00000000000000300000000000000… (on
4456 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
4457 input format. The maximum integer that will be printed in the new
4458 format is 8191 (four digits), while the old format always used a very
4459 long number (with the length varying by architecture), so they can be
4460 unambiguously distinguished.
4462 * /usr/sbin/halt.local is no longer supported. Implementation in
4463 distributions was inconsistent and it seems this functionality was
4466 To replace this functionality, users should:
4467 - either define a new unit and make it a dependency of final.target
4468 (systemctl add-wants final.target my-halt-local.service)
4469 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
4470 and ensure that it accepts "halt", "poweroff", "reboot", and
4471 "kexec" as an argument, see the description in systemd-shutdown(8).
4473 * When a [Match] section in .link or .network file is empty (contains
4474 no match patterns), a warning will be emitted. Please add any "match
4475 all" pattern instead, e.g. OriginalName=* or Name=* in case all
4476 interfaces should really be matched.
4478 * A new setting NUMAPolicy= may be used to set process memory
4479 allocation policy. This setting can be specified in
4480 /etc/systemd/system.conf and hence will set the default policy for
4481 PID1. The default policy can be overridden on a per-service
4482 basis. The related setting NUMAMask= is used to specify NUMA node
4483 mask that should be associated with the selected policy.
4485 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
4486 generates when processes it manages are reaching their memory limits,
4487 and will place their units in a special state, and optionally kill or
4488 stop the whole unit.
4490 * The service manager will now expose bus properties for the IO
4491 resources used by units. This information is also shown in "systemctl
4492 status" now (for services that have IOAccounting=yes set). Moreover,
4493 the IO accounting data is included in the resource log message
4494 generated whenever a unit stops.
4496 * Units may now configure an explicit timeout to wait for when killed
4497 with SIGABRT, for example when a service watchdog is hit. Previously,
4498 the regular TimeoutStopSec= timeout was applied in this case too —
4499 now a separate timeout may be set using TimeoutAbortSec=.
4501 * Services may now send a special WATCHDOG=trigger message with
4502 sd_notify() to trigger an immediate "watchdog missed" event, and thus
4503 trigger service termination. This is useful both for testing watchdog
4504 handling, but also for defining error paths in services, that shall
4505 be handled the same way as watchdog events.
4507 * There are two new per-unit settings IPIngressFilterPath= and
4508 IPEgressFilterPath= which allow configuration of a BPF program
4509 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
4510 to apply to the IP packet ingress/egress path of all processes of a
4511 unit. This is useful to allow running systemd services with BPF
4512 programs set up externally.
4514 * systemctl gained a new "clean" verb for removing the state, cache,
4515 runtime or logs directories of a service while it is terminated. The
4516 new verb may also be used to remove the state maintained on disk for
4517 timer units that have Persistent= configured.
4519 * During the last phase of shutdown systemd will now automatically
4520 increase the log level configured in the "kernel.printk" sysctl so
4521 that any relevant loggable events happening during late shutdown are
4522 made visible. Previously, loggable events happening so late during
4523 shutdown were generally lost if the "kernel.printk" sysctl was set to
4524 high thresholds, as regular logging daemons are terminated at that
4525 time and thus nothing is written to disk.
4527 * If processes terminated during the last phase of shutdown do not exit
4528 quickly systemd will now show their names after a short time, to make
4529 debugging easier. After a longer timeout they are forcibly killed,
4532 * journalctl (and the other tools that display logs) will now highlight
4533 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
4534 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
4535 are now shown in blue color, to separate them visually from regular
4536 logs. References to configuration files are now turned into clickable
4537 links on terminals that support that.
4539 * systemd-journald will now stop logging to /var/log/journal during
4540 shutdown when /var/ is on a separate mount, so that it can be
4541 unmounted safely during shutdown.
4543 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
4545 * systemd-resolved "Cache=" configuration option in resolved.conf has
4546 been extended to also accept the 'no-negative' value. Previously,
4547 only a boolean option was allowed (yes/no), having yes as the
4548 default. If this option is set to 'no-negative', negative answers are
4549 not cached while the old cache heuristics are used positive answers.
4550 The default remains unchanged.
4552 * The predictable naming scheme for network devices now supports
4553 generating predictable names for "netdevsim" devices.
4555 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
4558 Those two changes form a new net.naming-policy-scheme= entry.
4559 Distributions which want to preserve naming stability may want to set
4560 the -Ddefault-net-naming-scheme= configuration option.
4562 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
4563 interfaces natively.
4565 * systemd-networkd's bridge FDB support now allows configuration of a
4566 destination address for each entry (Destination=), as well as the
4567 VXLAN VNI (VNI=), as well as an option to declare what an entry is
4568 associated with (AssociatedWith=).
4570 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
4571 option for configuring the maximum number of DHCP lease requests. It
4572 also learnt a new BlackList= option for deny-listing DHCP servers (a
4573 similar setting has also been added to the IPv6 RA client), as well
4574 as a SendRelease= option for configuring whether to send a DHCP
4575 RELEASE message when terminating.
4577 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
4578 separately in the [DHCPv4] and [DHCPv6] sections.
4580 * systemd-networkd's DHCP support will now optionally create an
4581 implicit host route to the DNS server specified in the DHCP lease, in
4582 addition to the routes listed explicitly in the lease. This should
4583 ensure that in multi-homed systems DNS traffic leaves the systems on
4584 the interface that acquired the DNS server information even if other
4585 routes such as default routes exist. This behaviour may be turned on
4586 with the new RoutesToDNS= option.
4588 * systemd-networkd's VXLAN support gained a new option
4589 GenericProtocolExtension= for enabling VXLAN Generic Protocol
4590 Extension support, as well as IPDoNotFragment= for setting the IP
4591 "Don't fragment" bit on outgoing packets. A similar option has been
4592 added to the GENEVE support.
4594 * In systemd-networkd's [Route] section you may now configure
4595 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
4596 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
4597 propagation. The Type= setting now supports local, broadcast,
4598 anycast, multicast, any, xresolve routes, too.
4600 * systemd-networkd's [Network] section learnt a new option
4601 DefaultRouteOnDevice= for automatically configuring a default route
4602 onto the network device.
4604 * systemd-networkd's bridging support gained two new options ProxyARP=
4605 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
4606 MulticastRouter= for configuring multicast routing behaviour. A new
4607 option MulticastIGMPVersion= may be used to change bridge's multicast
4608 Internet Group Management Protocol (IGMP) version.
4610 * systemd-networkd's FooOverUDP support gained the ability to configure
4611 local and peer IP addresses via Local= and Peer=. A new option
4612 PeerPort= may be used to configure the peer's IP port.
4614 * systemd-networkd's TUN support gained a new setting VnetHeader= for
4615 tweaking Generic Segment Offload support.
4617 * The address family for policy rules may be specified using the new
4618 Family= option in the [RoutingPolicyRule] section.
4620 * networkctl gained a new "delete" command for removing virtual network
4621 devices, as well as a new "--stats" switch for showing device
4624 * networkd.conf gained a new setting SpeedMeter= and
4625 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
4626 measured speed may be shown by 'networkctl status'.
4628 * "networkctl status" now displays MTU and queue lengths, and more
4629 detailed information about VXLAN and bridge devices.
4631 * systemd-networkd's .network and .link files gained a new Property=
4632 setting in the [Match] section, to match against devices with
4633 specific udev properties.
4635 * systemd-networkd's tunnel support gained a new option
4636 AssignToLoopback= for selecting whether to use the loopback device
4637 "lo" as underlying device.
4639 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
4640 been renamed to LinkLayerAddress=, and it now allows configuration of
4643 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
4644 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
4645 IPv6 configuration (static or DHCPv6) was found for a given
4646 interface. It will not touch the sysctl otherwise.
4648 * The order of entries is $PATH used by the user manager instance was
4649 changed to put bin/ entries before the corresponding sbin/ entries.
4650 It is recommended to not rely on this order, and only ever have one
4651 binary with a given name in the system paths under /usr.
4653 * A new tool systemd-network-generator has been added that may generate
4654 .network, .netdev and .link files from IP configuration specified on
4655 the kernel command line in the format used by Dracut.
4657 * The CriticalConnection= setting in .network files is now deprecated,
4658 and replaced by a new KeepConfiguration= setting which allows more
4659 detailed configuration of the IP configuration to keep in place.
4661 * systemd-analyze gained a few new verbs:
4663 - "systemd-analyze timestamp" parses and converts timestamps. This is
4664 similar to the existing "systemd-analyze calendar" command which
4665 does the same for recurring calendar events.
4667 - "systemd-analyze timespan" parses and converts timespans (i.e.
4668 durations as opposed to points in time).
4670 - "systemd-analyze condition" will parse and test ConditionXYZ=
4673 - "systemd-analyze exit-status" will parse and convert exit status
4674 codes to their names and back.
4676 - "systemd-analyze unit-files" will print a list of all unit
4677 file paths and unit aliases.
4679 * SuccessExitStatus=, RestartPreventExitStatus=, and
4680 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
4681 is equivalent to "65"). Those exit status name mappings may be
4682 displayed with the systemd-analyze exit-status verb describe above.
4684 * systemd-logind now exposes a per-session SetBrightness() bus call,
4685 which may be used to securely change the brightness of a kernel
4686 brightness device, if it belongs to the session's seat. By using this
4687 call unprivileged clients can make changes to "backlight" and "leds"
4688 devices securely with strict requirements on session membership.
4689 Desktop environments may use this to generically make brightness
4690 changes to such devices without shipping private SUID binaries or
4691 udev rules for that purpose.
4693 * "udevadm info" gained a --wait-for-initialization switch to wait for
4694 a device to be initialized.
4696 * systemd-hibernate-resume-generator will now look for resumeflags= on
4697 the kernel command line, which is similar to rootflags= and may be
4698 used to configure device timeout for the hibernation device.
4700 * sd-event learnt a new API call sd_event_source_disable_unref() for
4701 disabling and unref'ing an event source in a single function. A
4702 related call sd_event_source_disable_unrefp() has been added for use
4703 with gcc's cleanup extension.
4705 * The sd-id128.h public API gained a new definition
4706 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
4709 * "busctl introspect" gained a new switch --xml-interface for dumping
4710 XML introspection data unmodified.
4712 * PID 1 may now show the unit name instead of the unit description
4713 string in its status output during boot. This may be configured in
4714 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
4715 kernel command line option systemd.status_unit_format=.
4717 * PID 1 now understands a new option KExecWatchdogSec= in
4718 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
4719 Previously watchdog functionality was only available for regular
4720 reboots. The new setting defaults to off, because we don't know in
4721 the general case if the watchdog will be reset after kexec (some
4722 drivers do reset it, but not all), and the new userspace might not be
4723 configured to handle the watchdog.
4725 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
4726 RebootWatchdogSec= to more clearly communicate what it is about. The
4727 old name is still accepted for compatibility.
4729 * The systemd.debug_shell kernel command line option now optionally
4730 takes a tty name to spawn the debug shell on, which allows a
4731 different tty to be selected than the built-in default.
4733 * Service units gained a new ExecCondition= setting which will run
4734 before ExecStartPre= and either continue execution of the unit (for
4735 clean exit codes), stop execution without marking the unit failed
4736 (for exit codes 1 through 254), or stop execution and fail the unit
4737 (for exit code 255 or abnormal termination).
4739 * A new service systemd-pstore.service has been added that pulls data
4740 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
4743 * timedatectl gained new verbs for configuring per-interface NTP
4744 service configuration for systemd-timesyncd.
4746 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
4747 2019. (You can set non-UTF-8 locales though, if you know their name.)
4749 * If variable assignments in sysctl.d/ files are prefixed with "-" any
4750 failures to apply them are now ignored.
4752 * systemd-random-seed.service now optionally credits entropy when
4753 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
4754 true for the service to enable this behaviour, but please consult the
4755 documentation first, since this comes with a couple of caveats.
4757 * systemd-random-seed.service is now a synchronization point for full
4758 initialization of the kernel's entropy pool. Services that require
4759 /dev/urandom to be correctly initialized should be ordered after this
4762 * The systemd-boot boot loader has been updated to optionally maintain
4763 a random seed file in the EFI System Partition (ESP). During the boot
4764 phase, this random seed is read and updated with a new seed
4765 cryptographically derived from it. Another derived seed is passed to
4766 the OS. The latter seed is then credited to the kernel's entropy pool
4767 very early during userspace initialization (from PID 1). This allows
4768 systems to boot up with a fully initialized kernel entropy pool from
4769 earliest boot on, and thus entirely removes all entropy pool
4770 initialization delays from systems using systemd-boot. Special care
4771 is taken to ensure different seeds are derived on system images
4772 replicated to multiple systems. "bootctl status" will show whether
4773 a seed was received from the boot loader.
4775 * bootctl gained two new verbs:
4777 - "bootctl random-seed" will generate the file in ESP and an EFI
4778 variable to allow a random seed to be passed to the OS as described
4781 - "bootctl is-installed" checks whether systemd-boot is currently
4784 * bootctl will warn if it detects that boot entries are misconfigured
4785 (for example if the kernel image was removed without purging the
4788 * A new document has been added describing systemd's use and support
4789 for the kernel's entropy pool subsystem:
4791 https://systemd.io/RANDOM_SEEDS
4793 * When the system is hibernated the swap device to write the
4794 hibernation image to is now automatically picked from all available
4795 swap devices, preferring the swap device with the highest configured
4796 priority over all others, and picking the device with the most free
4797 space if there are multiple devices with the highest priority.
4799 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
4800 option that permits selecting the timeout how long to wait for a
4801 device with an encryption key before asking for the password.
4803 * IOWeight= has learnt to properly set the IO weight when using the
4804 BFQ scheduler officially found in kernels 5.0+.
4806 * A new mailing list has been created for reporting of security issues:
4807 systemd-security@redhat.com. For mode details, see
4808 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
4810 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
4811 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
4812 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
4813 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
4814 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
4815 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
4816 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
4817 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
4818 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
4819 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
4820 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
4821 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
4822 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
4823 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
4824 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
4825 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
4826 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
4827 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
4828 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
4829 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
4830 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
4831 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
4832 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
4833 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
4834 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
4835 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
4836 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
4837 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
4838 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
4839 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
4841 – Camerino, 2019-09-03
4845 * In .link files, MACAddressPolicy=persistent (the default) is changed
4846 to cover more devices. For devices like bridges, tun, tap, bond, and
4847 similar interfaces that do not have other identifying information,
4848 the interface name is used as the basis for persistent seed for MAC
4849 and IPv4LL addresses. The way that devices that were handled
4850 previously is not changed, and this change is about covering more
4851 devices then previously by the "persistent" policy.
4853 MACAddressPolicy=random may be used to force randomized MACs and
4854 IPv4LL addresses for a device if desired.
4856 Hint: the log output from udev (at debug level) was enhanced to
4857 clarify what policy is followed and which attributes are used.
4858 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
4859 may be used to view this.
4861 Hint: if a bridge interface is created without any slaves, and gains
4862 a slave later, then now the bridge does not inherit slave's MAC.
4863 To inherit slave's MAC, for example, create the following file:
4865 # /etc/systemd/network/98-bridge-inherit-mac.link
4870 MACAddressPolicy=none
4873 * The .device units generated by systemd-fstab-generator and other
4874 generators do not automatically pull in the corresponding .mount unit
4875 as a Wants= dependency. This means that simply plugging in the device
4876 will not cause the mount unit to be started automatically. But please
4877 note that the mount unit may be started for other reasons, in
4878 particular if it is part of local-fs.target, and any unit which
4879 (transitively) depends on local-fs.target is started.
4881 * networkctl list/status/lldp now accept globbing wildcards for network
4882 interface names to match against all existing interfaces.
4884 * The $PIDFILE environment variable is set to point the absolute path
4885 configured with PIDFile= for processes of that service.
4887 * The fallback DNS server list was augmented with Cloudflare public DNS
4888 servers. Use `-Ddns-servers=` to set a different fallback.
4890 * A new special target usb-gadget.target will be started automatically
4891 when a USB Device Controller is detected (which means that the system
4892 is a USB peripheral).
4894 * A new unit setting CPUQuotaPeriodSec= assigns the time period
4895 relatively to which the CPU time quota specified by CPUQuota= is
4898 * A new unit setting ProtectHostname= may be used to prevent services
4899 from modifying hostname information (even if they otherwise would
4900 have privileges to do so).
4902 * A new unit setting NetworkNamespacePath= may be used to specify a
4903 namespace for service or socket units through a path referring to a
4904 Linux network namespace pseudo-file.
4906 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
4907 have an effect on .socket units: when used the listening socket is
4908 created within the configured network namespace instead of the host
4911 * ExecStart= command lines in unit files may now be prefixed with ':'
4912 in which case environment variable substitution is
4913 disabled. (Supported for the other ExecXYZ= settings, too.)
4915 * .timer units gained two new boolean settings OnClockChange= and
4916 OnTimezoneChange= which may be used to also trigger a unit when the
4917 system clock is changed or the local timezone is
4918 modified. systemd-run has been updated to make these options easily
4919 accessible from the command line for transient timers.
4921 * Two new conditions for units have been added: ConditionMemory= may be
4922 used to conditionalize a unit based on installed system
4923 RAM. ConditionCPUs= may be used to conditionalize a unit based on
4924 installed CPU cores.
4926 * The @default system call filter group understood by SystemCallFilter=
4927 has been updated to include the new rseq() system call introduced in
4930 * A new time-set.target has been added that indicates that the system
4931 time has been set from a local source (possibly imprecise). The
4932 existing time-sync.target is stronger and indicates that the time has
4933 been synchronized with a precise external source. Services where
4934 approximate time is sufficient should use the new target.
4936 * "systemctl start" (and related commands) learnt a new
4937 --show-transaction option. If specified brief information about all
4938 jobs queued because of the requested operation is shown.
4940 * systemd-networkd recognizes a new operation state 'enslaved', used
4941 (instead of 'degraded' or 'carrier') for interfaces which form a
4942 bridge, bond, or similar, and an new 'degraded-carrier' operational
4943 state used for the bond or bridge master interface when one of the
4944 enslaved devices is not operational.
4946 * .network files learnt the new IgnoreCarrierLoss= option for leaving
4947 networks configured even if the carrier is lost.
4949 * The RequiredForOnline= setting in .network files may now specify a
4950 minimum operational state required for the interface to be considered
4951 "online" by systemd-networkd-wait-online. Related to this
4952 systemd-networkd-wait-online gained a new option --operational-state=
4953 to configure the same, and its --interface= option was updated to
4954 optionally also take an operational state specific for an interface.
4956 * systemd-networkd-wait-online gained a new setting --any for waiting
4957 for only one of the requested interfaces instead of all of them.
4959 * systemd-networkd now implements L2TP tunnels.
4961 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
4962 may be used to cause autonomous and onlink prefixes received in IPv6
4963 Router Advertisements to be ignored.
4965 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
4966 file settings may be used to tweak bridge behaviour.
4968 * The new TripleSampling= option in .network files may be used to
4969 configure CAN triple sampling.
4971 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
4972 used to point to private or preshared key for a WireGuard interface.
4974 * /etc/crypttab now supports the same-cpu-crypt and
4975 submit-from-crypt-cpus options to tweak encryption work scheduling
4978 * systemd-tmpfiles will now take a BSD file lock before operating on a
4979 contents of directory. This may be used to temporarily exclude
4980 directories from aging by taking the same lock (useful for example
4981 when extracting a tarball into /tmp or /var/tmp as a privileged user,
4982 which might create files with really old timestamps, which
4983 nevertheless should not be deleted). For further details, see:
4985 https://systemd.io/TEMPORARY_DIRECTORIES
4987 * systemd-tmpfiles' h line type gained support for the
4988 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
4989 controlling project quota inheritance.
4991 * sd-boot and bootctl now implement support for an Extended Boot Loader
4992 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
4993 addition to the ESP partition mounted to /efi or /boot/efi.
4994 Configuration file fragments, kernels, initrds and other EFI images
4995 to boot will be loaded from both the ESP and XBOOTLDR partitions.
4996 The XBOOTLDR partition was previously described by the Boot Loader
4997 Specification, but implementation was missing in sd-boot. Support for
4998 this concept allows using the sd-boot boot loader in more
4999 conservative scenarios where the boot loader itself is placed in the
5000 ESP but the kernels to boot (and their metadata) in a separate
5003 * A system may now be booted with systemd.volatile=overlay on the
5004 kernel command line, which causes the root file system to be set up
5005 an overlayfs mount combining the root-only root directory with a
5006 writable tmpfs. In this setup, the underlying root device is not
5007 modified, and any changes are lost at reboot.
5009 * Similar, systemd-nspawn can now boot containers with a volatile
5010 overlayfs root with the new --volatile=overlay switch.
5012 * systemd-nspawn can now consume OCI runtime bundles using a new
5013 --oci-bundle= option. This implementation is fully usable, with most
5014 features in the specification implemented, but since this a lot of
5015 new code and functionality, this feature should most likely not
5016 be used in production yet.
5018 * systemd-nspawn now supports various options described by the OCI
5019 runtime specification on the command-line and in .nspawn files:
5020 --inaccessible=/Inaccessible= may be used to mask parts of the file
5021 system tree, --console=/--pipe may be used to configure how standard
5022 input, output, and error are set up.
5024 * busctl learned the `emit` verb to generate D-Bus signals.
5026 * systemd-analyze cat-config may be used to gather and display
5027 configuration spread over multiple files, for example system and user
5028 presets, tmpfiles.d, sysusers.d, udev rules, etc.
5030 * systemd-analyze calendar now takes an optional new parameter
5031 --iterations= which may be used to show a maximum number of iterations
5032 the specified expression will elapse next.
5034 * The sd-bus C API gained support for naming method parameters in the
5037 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
5038 the reboot() system call expects.
5040 * journalctl learnt a new --cursor-file= option that points to a file
5041 from which a cursor should be loaded in the beginning and to which
5042 the updated cursor should be stored at the end.
5044 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
5045 detected by systemd-detect-virt (and may also be used in
5046 ConditionVirtualization=).
5048 * The behaviour of systemd-logind may now be modified with environment
5049 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
5050 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
5051 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
5052 skip the relevant operation completely (when set to false), or to
5053 create a flag file in /run/systemd (when set to true), instead of
5054 actually commencing the real operation when requested. The presence
5055 of /run/systemd/reboot-to-firmware-setup,
5056 /run/systemd/reboot-to-boot-loader-menu, and
5057 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
5058 boot loader implementations to replace some steps logind performs
5059 during reboot with their own operations.
5061 * systemctl can be used to request a reboot into the boot loader menu
5062 or a specific boot loader entry with the new --boot-load-menu= and
5063 --boot-loader-entry= options to a reboot command. (This requires a
5064 boot loader that supports this, for example sd-boot.)
5066 * kernel-install will no longer unconditionally create the output
5067 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
5068 snippets, but will do only if the machine-specific parent directory
5069 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
5070 to create this parent directory during sd-boot installation.
5072 This makes it easier to use kernel-install with plugins which support
5073 a different layout of the bootloader partitions (for example grub2).
5075 * During package installation (with `ninja install`), we would create
5076 symlinks for getty@tty1.service, systemd-networkd.service,
5077 systemd-networkd.socket, systemd-resolved.service,
5078 remote-cryptsetup.target, remote-fs.target,
5079 systemd-networkd-wait-online.service, and systemd-timesyncd.service
5080 in /etc, as if `systemctl enable` was called for those units, to make
5081 the system usable immediately after installation. Now this is not
5082 done anymore, and instead calling `systemctl preset-all` is
5083 recommended after the first installation of systemd.
5085 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
5086 is built on seccomp. When turned on creation of SUID/SGID files is
5089 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
5090 implied if DynamicUser= is turned on for a service. This hardens
5091 these services, so that they neither can benefit from nor create
5092 SUID/SGID executables. This is a minor compatibility breakage, given
5093 that when DynamicUser= was first introduced SUID/SGID behaviour was
5094 unaffected. However, the security benefit of these two options is
5095 substantial, and the setting is still relatively new, hence we opted
5096 to make it mandatory for services with dynamic users.
5098 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
5099 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
5100 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
5101 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
5102 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
5103 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
5104 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
5105 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
5106 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
5107 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
5108 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
5109 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
5110 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
5111 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
5112 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
5113 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
5114 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
5115 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5117 — Warsaw, 2019-04-11
5121 * The default locale can now be configured at compile time. Otherwise,
5122 a suitable default will be selected automatically (one of C.UTF-8,
5123 en_US.UTF-8, and C).
5125 * The version string shown by systemd and other tools now includes the
5126 git commit hash when built from git. An override may be specified
5127 during compilation, which is intended to be used by distributions to
5128 include the package release information.
5130 * systemd-cat can now filter standard input and standard error streams
5131 for different syslog priorities using the new --stderr-priority=
5134 * systemd-journald and systemd-journal-remote reject entries which
5135 contain too many fields (CVE-2018-16865) and set limits on the
5136 process' command line length (CVE-2018-16864).
5138 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
5141 * A new network device NamePolicy "keep" is implemented for link files,
5142 and used by default in 99-default.link (the fallback configuration
5143 provided by systemd). With this policy, if the network device name
5144 was already set by userspace, the device will not be renamed again.
5145 This matches the naming scheme that was implemented before
5146 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
5147 is also enabled by default, even if not specified. Effectively, this
5148 means that if naming-scheme >= 240 is specified, network devices will
5149 be renamed according to the configuration, even if they have been
5150 renamed already, if "keep" is not specified as the naming policy in
5151 the .link file. The 99-default.link file provided by systemd includes
5152 "keep" for backwards compatibility, but it is recommended for user
5153 installed .link files to *not* include it.
5155 The "kernel" policy, which keeps kernel names declared to be
5156 "persistent", now works again as documented.
5158 * kernel-install script now optionally takes the paths to one or more
5159 initrd files, and passes them to all plugins.
5161 * The mincore() system call has been dropped from the @system-service
5162 system call filter group, as it is pretty exotic and may potentially
5163 used for side-channel attacks.
5165 * -fPIE is dropped from compiler and linker options. Please specify
5166 -Db_pie=true option to meson to build position-independent
5167 executables. Note that the meson option is supported since meson-0.49.
5169 * The fs.protected_regular and fs.protected_fifos sysctls, which were
5170 added in Linux 4.19 to make some data spoofing attacks harder, are
5171 now enabled by default. While this will hopefully improve the
5172 security of most installations, it is technically a backwards
5173 incompatible change; to disable these sysctls again, place the
5174 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
5176 fs.protected_regular = 0
5177 fs.protected_fifos = 0
5179 Note that the similar hardlink and symlink protection has been
5180 enabled since v199, and may be disabled likewise.
5182 * The files read from the EnvironmentFile= setting in unit files now
5183 parse backslashes inside quotes literally, matching the behaviour of
5186 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
5187 now automatically become NOPs when run in a chroot() environment.
5189 * The tmpfiles.d/ "C" line type will now copy directory trees not only
5190 when the destination is so far missing, but also if it already exists
5191 as a directory and is empty. This is useful to cater for systems
5192 where directory trees are put together from multiple separate mount
5193 points but otherwise empty.
5195 * A new function sd_bus_close_unref() (and the associated
5196 sd_bus_close_unrefp()) has been added to libsystemd, that combines
5197 sd_bus_close() and sd_bus_unref() in one.
5199 * udevadm control learnt a new option for --ping for testing whether a
5200 systemd-udevd instance is running and reacting.
5202 * udevadm trigger learnt a new option for --wait-daemon for waiting
5203 systemd-udevd daemon to be initialized.
5205 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
5206 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
5207 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
5208 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
5209 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
5210 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
5211 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
5212 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
5213 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
5214 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
5215 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
5216 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
5217 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
5218 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
5219 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
5220 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
5221 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
5223 — Berlin, 2019-02-14
5227 * NoNewPrivileges=yes has been set for all long-running services
5228 implemented by systemd. Previously, this was problematic due to
5229 SELinux (as this would also prohibit the transition from PID1's label
5230 to the service's label). This restriction has since been lifted, but
5231 an SELinux policy update is required.
5232 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
5234 * DynamicUser=yes is dropped from systemd-networkd.service,
5235 systemd-resolved.service and systemd-timesyncd.service, which was
5236 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
5237 and since v236 for systemd-timesyncd.service. The users and groups
5238 systemd-network, systemd-resolve and systemd-timesync are created
5239 by systemd-sysusers again. Distributors or system administrators
5240 may need to create these users and groups if they not exist (or need
5241 to re-enable DynamicUser= for those units) while upgrading systemd.
5242 Also, the clock file for systemd-timesyncd may need to move from
5243 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
5245 * When unit files are loaded from disk, previously systemd would
5246 sometimes (depending on the unit loading order) load units from the
5247 target path of symlinks in .wants/ or .requires/ directories of other
5248 units. This meant that unit could be loaded from different paths
5249 depending on whether the unit was requested explicitly or as a
5250 dependency of another unit, not honouring the priority of directories
5251 in search path. It also meant that it was possible to successfully
5252 load and start units which are not found in the unit search path, as
5253 long as they were requested as a dependency and linked to from
5254 .wants/ or .requires/. The target paths of those symlinks are not
5255 used for loading units anymore and the unit file must be found in
5258 * A new service type has been added: Type=exec. It's very similar to
5259 Type=simple but ensures the service manager will wait for both fork()
5260 and execve() of the main service binary to complete before proceeding
5261 with follow-up units. This is primarily useful so that the manager
5262 propagates any errors in the preparation phase of service execution
5263 back to the job that requested the unit to be started. For example,
5264 consider a service that has ExecStart= set to a file system binary
5265 that doesn't exist. With Type=simple starting the unit would be
5266 considered instantly successful, as only fork() has to complete
5267 successfully and the manager does not wait for execve(), and hence
5268 its failure is seen "too late". With the new Type=exec service type
5269 starting the unit will fail, as the manager will wait for the
5270 execve() and notice its failure, which is then propagated back to the
5273 NOTE: with the next release 241 of systemd we intend to change the
5274 systemd-run tool to default to Type=exec for transient services
5275 started by it. This should be mostly safe, but in specific corner
5276 cases might result in problems, as the systemd-run tool will then
5277 block on NSS calls (such as user name look-ups due to User=) done
5278 between the fork() and execve(), which under specific circumstances
5279 might cause problems. It is recommended to specify "-p Type=simple"
5280 explicitly in the few cases where this applies. For regular,
5281 non-transient services (i.e. those defined with unit files on disk)
5282 we will continue to default to Type=simple.
5284 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
5285 userspace processes is set to 1024 (soft) and 4096
5286 (hard). Previously, systemd passed this on unmodified to all
5287 processes it forked off. With this systemd release the hard limit
5288 systemd passes on is increased to 512K, overriding the kernel's
5289 defaults and substantially increasing the number of simultaneous file
5290 descriptors unprivileged userspace processes can allocate. Note that
5291 the soft limit remains at 1024 for compatibility reasons: the
5292 traditional UNIX select() call cannot deal with file descriptors >=
5293 1024 and increasing the soft limit globally might thus result in
5294 programs unexpectedly allocating a high file descriptor and thus
5295 failing abnormally when attempting to use it with select() (of
5296 course, programs shouldn't use select() anymore, and prefer
5297 poll()/epoll, but the call unfortunately remains undeservedly popular
5298 at this time). This change reflects the fact that file descriptor
5299 handling in the Linux kernel has been optimized in more recent
5300 kernels and allocating large numbers of them should be much cheaper
5301 both in memory and in performance than it used to be. Programs that
5302 want to take benefit of the increased limit have to "opt-in" into
5303 high file descriptors explicitly by raising their soft limit. Of
5304 course, when they do that they must acknowledge that they cannot use
5305 select() anymore (and neither can any shared library they use — or
5306 any shared library used by any shared library they use and so on).
5307 Which default hard limit is most appropriate is of course hard to
5308 decide. However, given reports that ~300K file descriptors are used
5309 in real-life applications we believe 512K is sufficiently high as new
5310 default for now. Note that there are also reports that using very
5311 high hard limits (e.g. 1G) is problematic: some software allocates
5312 large arrays with one element for each potential file descriptor
5313 (Java, …) — a high hard limit thus triggers excessively large memory
5314 allocations in these applications. Hopefully, the new default of 512K
5315 is a good middle ground: higher than what real-life applications
5316 currently need, and low enough for avoid triggering excessively large
5317 allocations in problematic software. (And yes, somebody should fix
5320 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
5321 to the highest possible values, as separate accounting of file
5322 descriptors is no longer necessary, as memcg tracks them correctly as
5323 part of the memory accounting anyway. Thus, from the four limits on
5324 file descriptors currently enforced (fs.file-max, fs.nr_open,
5325 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
5326 and keep only the latter two. A set of build-time options
5327 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
5328 has been added to revert this change in behaviour, which might be
5329 an option for systems that turn off memcg in the kernel.
5331 * When no /etc/locale.conf file exists (and hence no locale settings
5332 are in place), systemd will now use the "C.UTF-8" locale by default,
5333 and set LANG= to it. This locale is supported by various
5334 distributions including Fedora, with clear indications that upstream
5335 glibc is going to make it available too. This locale enables UTF-8
5336 mode by default, which appears appropriate for 2018.
5338 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
5339 default. This effectively switches the RFC3704 Reverse Path filtering
5340 from Strict mode to Loose mode. This is more appropriate for hosts
5341 that have multiple links with routes to the same networks (e.g.
5342 a client with a Wi-Fi and Ethernet both connected to the internet).
5344 Consult the kernel documentation for details on this sysctl:
5345 https://docs.kernel.org/networking/ip-sysctl.html
5347 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
5350 * CPUAccounting=yes no longer enables the CPU controller when using
5351 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
5352 statistics are now provided independently from the CPU controller.
5354 * Support for disabling a particular cgroup controller within a sub-tree
5355 has been added through the DisableControllers= directive.
5357 * cgroup_no_v1=all on the kernel command line now also implies
5358 using the unified cgroup hierarchy, unless one explicitly passes
5359 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
5361 * The new "MemoryMin=" unit file property may now be used to set the
5362 memory usage protection limit of processes invoked by the unit. This
5363 controls the cgroup v2 memory.min attribute. Similarly, the new
5364 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
5365 cgroup v2 io.latency cgroup property for configuring per-service I/O
5368 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
5369 to the cgroup v1 "devices" cgroup controller.
5371 * systemd-escape now is able to combine --unescape with --template. It
5372 also learnt a new option --instance for extracting and unescaping the
5373 instance part of a unit name.
5375 * sd-bus now provides the sd_bus_message_readv() which is similar to
5376 sd_bus_message_read() but takes a va_list object. The pair
5377 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
5378 has been added for configuring the default method call timeout to
5379 use. sd_bus_error_move() may be used to efficiently move the contents
5380 from one sd_bus_error structure to another, invalidating the
5381 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
5382 be used to control whether a bus connection object is automatically
5383 flushed when an sd-event loop is exited.
5385 * When processing classic BSD syslog log messages, journald will now
5386 save the original time-stamp string supplied in the new
5387 SYSLOG_TIMESTAMP= journal field. This permits consumers to
5388 reconstruct the original BSD syslog message more correctly.
5390 * StandardOutput=/StandardError= in service files gained support for
5391 new "append:…" parameters, for connecting STDOUT/STDERR of a service
5392 to a file, and appending to it.
5394 * The signal to use as last step of killing of unit processes is now
5395 configurable. Previously it was hard-coded to SIGKILL, which may now
5396 be overridden with the new KillSignal= setting. Note that this is the
5397 signal used when regular termination (i.e. SIGTERM) does not suffice.
5398 Similarly, the signal used when aborting a program in case of a
5399 watchdog timeout may now be configured too (WatchdogSignal=).
5401 * The XDG_SESSION_DESKTOP environment variable may now be configured in
5402 the pam_systemd argument line, using the new desktop= switch. This is
5403 useful to initialize it properly from a display manager without
5404 having to touch C code.
5406 * Most configuration options that previously accepted percentage values
5407 now also accept permille values with the '‰' suffix (instead of '%').
5409 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
5412 * systemd-resolved's configuration file resolved.conf gained a new
5413 option ReadEtcHosts= which may be used to turn off processing and
5414 honoring /etc/hosts entries.
5416 * The "--wait" switch may now be passed to "systemctl
5417 is-system-running", in which case the tool will synchronously wait
5418 until the system finished start-up.
5420 * hostnamed gained a new bus call to determine the DMI product UUID.
5422 * On x86-64 systemd will now prefer using the RDRAND processor
5423 instruction over /dev/urandom whenever it requires randomness that
5424 neither has to be crypto-grade nor should be reproducible. This
5425 should substantially reduce the amount of entropy systemd requests
5426 from the kernel during initialization on such systems, though not
5427 reduce it to zero. (Why not zero? systemd still needs to allocate
5428 UUIDs and such uniquely, which require high-quality randomness.)
5430 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
5431 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
5432 for forcing the "Other Information" bit in IPv6 RA messages. The
5433 bonding logic gained four new options AdActorSystemPriority=,
5434 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
5435 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
5436 shuffling of flows. The tunnel logic gained a new
5437 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
5438 Deployment. The policy rule logic gained four new options IPProtocol=,
5439 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
5440 support for the MulticastToUnicast= option. networkd also gained
5441 support for configuring static IPv4 ARP or IPv6 neighbor entries.
5443 * .preset files (as read by 'systemctl preset') may now be used to
5444 instantiate services.
5446 * /etc/crypttab now understands the sector-size= option to configure
5447 the sector size for an encrypted partition.
5449 * Key material for encrypted disks may now be placed on a formatted
5450 medium, and referenced from /etc/crypttab by the UUID of the file
5451 system, followed by "=" suffixed by the path to the key file.
5453 * The "collect" udev component has been removed without replacement, as
5454 it is neither used nor maintained.
5456 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
5457 LogsDirectory=, ConfigurationDirectory= settings are used in a
5458 service the executed processes will now receive a set of environment
5459 variables containing the full paths of these directories.
5460 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
5461 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
5462 are used. Note that these options may be used multiple times per
5463 service in which case the resulting paths will be concatenated and
5464 separated by colons.
5466 * Predictable interface naming has been extended to cover InfiniBand
5467 NICs. They will be exposed with an "ib" prefix.
5469 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
5470 which case the respective line failing is ignored.
5472 * .link files may now be used to configure the equivalent to the
5473 "ethtool advertise" commands.
5475 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
5476 alternative to libudev.h. Previously, the latter was just an internal
5477 wrapper around the former, but now these two APIs are exposed
5480 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
5481 which calculates an app-specific boot ID similar to how
5482 sd_id128_get_machine_app_specific() generates an app-specific machine
5485 * A new tool systemd-id128 has been added that can be used to determine
5486 and generate various 128bit IDs.
5488 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
5491 * systemd-hibernate-resume-generator will now honor the "noresume"
5492 kernel command line option, in which case it will bypass resuming
5493 from any hibernated image.
5495 * The systemd-sleep.conf configuration file gained new options
5496 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
5497 AllowHybridSleep= for prohibiting specific sleep modes even if the
5498 kernel exports them.
5500 * portablectl is now officially supported and has thus moved to
5503 * bootctl learnt the two new commands "set-default" and "set-oneshot"
5504 for setting the default boot loader item to boot to (either
5505 persistently or only for the next boot). This is currently only
5506 compatible with sd-boot, but may be implemented on other boot loaders
5507 too, that follow the boot loader interface. The updated interface is
5508 now documented here:
5510 https://systemd.io/BOOT_LOADER_INTERFACE
5512 * A new kernel command line option systemd.early_core_pattern= is now
5513 understood which may be used to influence the core_pattern PID 1
5514 installs during early boot.
5516 * busctl learnt two new options -j and --json= for outputting method
5517 call replies, properties and monitoring output in JSON.
5519 * journalctl's JSON output now supports simple ANSI coloring as well as
5520 a new "json-seq" mode for generating RFC7464 output.
5522 * Unit files now support the %g/%G specifiers that resolve to the UNIX
5523 group/GID of the service manager runs as, similar to the existing
5524 %u/%U specifiers that resolve to the UNIX user/UID.
5526 * systemd-logind learnt a new global configuration option
5527 UserStopDelaySec= that may be set in logind.conf. It specifies how
5528 long the systemd --user instance shall remain started after a user
5529 logs out. This is useful to speed up repetitive re-connections of the
5530 same user, as it means the user's service manager doesn't have to be
5531 stopped/restarted on each iteration, but can be reused between
5532 subsequent options. This setting defaults to 10s. systemd-logind also
5533 exports two new properties on its Manager D-Bus objects indicating
5534 whether the system's lid is currently closed, and whether the system
5537 * systemd gained support for a generic boot counting logic, which
5538 generically permits automatic reverting to older boot loader entries
5539 if newer updated ones don't work. The boot loader side is implemented
5540 in sd-boot, but is kept open for other boot loaders too. For details
5543 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
5545 * The SuccessAction=/FailureAction= unit file settings now learnt two
5546 new parameters: "exit" and "exit-force", which result in immediate
5547 exiting of the service manager, and are only useful in systemd --user
5548 and container environments.
5550 * Unit files gained support for a pair of options
5551 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
5552 exit status to use as service manager exit status when
5553 SuccessAction=/FailureAction= is set to exit or exit-force.
5555 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
5556 options may now be used to configure the log rate limiting applied by
5557 journald per-service.
5559 * systemd-analyze gained a new verb "timespan" for parsing and
5560 normalizing time span values (i.e. strings like "5min 7s 8us").
5562 * systemd-analyze also gained a new verb "security" for analyzing the
5563 security and sand-boxing settings of services in order to determine an
5564 "exposure level" for them, indicating whether a service would benefit
5565 from more sand-boxing options turned on for them.
5567 * "systemd-analyze syscall-filter" will now also show system calls
5568 supported by the local kernel but not included in any of the defined
5571 * .nspawn files now understand the Ephemeral= setting, matching the
5572 --ephemeral command line switch.
5574 * sd-event gained the new APIs sd_event_source_get_floating() and
5575 sd_event_source_set_floating() for controlling whether a specific
5576 event source is "floating", i.e. destroyed along with the even loop
5579 * Unit objects on D-Bus gained a new "Refs" property that lists all
5580 clients that currently have a reference on the unit (to ensure it is
5583 * The JoinControllers= option in system.conf is no longer supported, as
5584 it didn't work correctly, is hard to support properly, is legacy (as
5585 the concept only exists on cgroup v1) and apparently wasn't used.
5587 * Journal messages that are generated whenever a unit enters the failed
5588 state are now tagged with a unique MESSAGE_ID. Similarly, messages
5589 generated whenever a service process exits are now made recognizable,
5590 too. A tagged message is also emitted whenever a unit enters the
5591 "dead" state on success.
5593 * systemd-run gained a new switch --working-directory= for configuring
5594 the working directory of the service to start. A shortcut -d is
5595 equivalent, setting the working directory of the service to the
5596 current working directory of the invoking program. The new --shell
5597 (or just -S) option has been added for invoking the $SHELL of the
5598 caller as a service, and implies --pty --same-dir --wait --collect
5599 --service-type=exec. Or in other words, "systemd-run -S" is now the
5600 quickest way to quickly get an interactive in a fully clean and
5601 well-defined system service context.
5603 * machinectl gained a new verb "import-fs" for importing an OS tree
5604 from a directory. Moreover, when a directory or tarball is imported
5605 and single top-level directory found with the OS itself below the OS
5606 tree is automatically mangled and moved one level up.
5608 * systemd-importd will no longer set up an implicit btrfs loop-back
5609 file system on /var/lib/machines. If one is already set up, it will
5610 continue to be used.
5612 * A new generator "systemd-run-generator" has been added. It will
5613 synthesize a unit from one or more program command lines included in
5614 the kernel command line. This is very useful in container managers
5617 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
5619 This will run "systemd-nspawn" on an image, invoke the specified
5620 command line and immediately shut down the container again, returning
5621 the command line's exit code.
5623 * The block device locking logic is now documented:
5625 https://systemd.io/BLOCK_DEVICE_LOCKING
5627 * loginctl and machinectl now optionally output the various tables in
5628 JSON using the --output= switch. It is our intention to add similar
5629 support to systemctl and all other commands.
5631 * udevadm's query and trigger verb now optionally take a .device unit
5634 * systemd-udevd's network naming logic now understands a new
5635 net.naming-scheme= kernel command line switch, which may be used to
5636 pick a specific version of the naming scheme. This helps stabilizing
5637 interface names even as systemd/udev are updated and the naming logic
5640 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
5641 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
5642 initialize one to all 0xFF.
5644 * After loading the SELinux policy systemd will now recursively relabel
5645 all files and directories listed in
5646 /run/systemd/relabel-extra.d/*.relabel (which should be simple
5647 newline separated lists of paths) in addition to the ones it already
5648 implicitly relabels in /run, /dev and /sys. After the relabelling is
5649 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
5650 removed. This is useful to permit initrds (i.e. code running before
5651 the SELinux policy is in effect) to generate files in the host
5652 filesystem safely and ensure that the correct label is applied during
5653 the transition to the host OS.
5655 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
5656 mknod() handling in user namespaces. Previously mknod() would always
5657 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
5658 but device nodes generated that way cannot be opened, and attempts to
5659 open them result in EPERM. This breaks the "graceful fallback" logic
5660 in systemd's PrivateDevices= sand-boxing option. This option is
5661 implemented defensively, so that when systemd detects it runs in a
5662 restricted environment (such as a user namespace, or an environment
5663 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
5664 where device nodes cannot be created the effect of PrivateDevices= is
5665 bypassed (following the logic that 2nd-level sand-boxing is not
5666 essential if the system systemd runs in is itself already sand-boxed
5667 as a whole). This logic breaks with 4.18 in container managers where
5668 user namespacing is used: suddenly PrivateDevices= succeeds setting
5669 up a private /dev/ file system containing devices nodes — but when
5670 these are opened they don't work.
5672 At this point it is recommended that container managers utilizing
5673 user namespaces that intend to run systemd in the payload explicitly
5674 block mknod() with seccomp or similar, so that the graceful fallback
5677 We are very sorry for the breakage and the requirement to change
5678 container configurations for newer kernels. It's purely caused by an
5679 incompatible kernel change. The relevant kernel developers have been
5680 notified about this userspace breakage quickly, but they chose to
5683 * PermissionsStartOnly= setting is deprecated (but is still supported
5684 for backwards compatibility). The same functionality is provided by
5685 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
5688 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
5689 pam_systemd anymore.
5691 * The naming scheme for network devices was changed to always rename
5692 devices, even if they were already renamed by userspace. The "kernel"
5693 policy was changed to only apply as a fallback, if no other naming
5696 * The requirements to build systemd is bumped to meson-0.46 and
5699 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
5700 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
5701 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
5702 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
5703 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
5704 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
5705 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
5706 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
5707 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
5708 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
5709 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
5710 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
5711 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
5712 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
5713 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
5714 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
5715 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
5716 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
5717 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
5718 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
5719 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
5720 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
5721 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
5722 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
5723 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
5724 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
5725 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
5726 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
5727 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
5728 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
5729 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
5730 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
5731 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
5732 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
5733 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
5734 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
5735 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
5736 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
5737 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
5738 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
5739 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
5740 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
5741 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
5742 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
5743 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
5745 — Warsaw, 2018-12-21
5749 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
5750 builtin will name network interfaces differently than in previous
5751 versions for virtual network interfaces created with SR-IOV and NPAR
5752 and for devices where the PCI network controller device does not have
5753 a slot number associated.
5755 SR-IOV virtual devices are now named based on the name of the parent
5756 interface, with a suffix of "v<N>", where <N> is the virtual device
5757 number. Previously those virtual devices were named as if completely
5760 The ninth and later NPAR virtual devices will be named following the
5761 scheme used for the first eight NPAR partitions. Previously those
5762 devices were not renamed and the kernel default (eth<n>) was used.
5764 "net_id" will also generate names for PCI devices where the PCI
5765 network controller device does not have an associated slot number
5766 itself, but one of its parents does. Previously those devices were
5767 not renamed and the kernel default (eth<n>) was used.
5769 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
5770 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
5771 the unit. So, it is expected that the default behavior of
5772 systemd-logind is not changed. However, if distribution packagers or
5773 administrators disabled or modified IPAddressDeny= setting by a
5774 drop-in config file, then it may be necessary to update the file to
5775 re-enable AF_INET and AF_INET6 to support network user name services,
5778 * When the RestrictNamespaces= unit property is specified multiple
5779 times, then the specified types are merged now. Previously, only the
5780 last assignment was used. So, if distribution packagers or
5781 administrators modified the setting by a drop-in config file, then it
5782 may be necessary to update the file.
5784 * When OnFailure= is used in combination with Restart= on a service
5785 unit, then the specified units will no longer be triggered on
5786 failures that result in restarting. Previously, the specified units
5787 would be activated each time the unit failed, even when the unit was
5788 going to be restarted automatically. This behaviour contradicted the
5789 documentation. With this release the code is adjusted to match the
5792 * systemd-tmpfiles will now print a notice whenever it encounters
5793 tmpfiles.d/ lines referencing the /var/run/ directory. It will
5794 recommend reworking them to use the /run/ directory instead (for
5795 which /var/run/ is simply a symlinked compatibility alias). This way
5796 systemd-tmpfiles can properly detect line conflicts and merge lines
5797 referencing the same file by two paths, without having to access
5800 * systemctl disable/unmask/preset/preset-all cannot be used with
5801 --runtime. Previously this was allowed, but resulted in unintuitive
5802 behaviour that wasn't useful. systemctl disable/unmask will now undo
5803 both runtime and persistent enablement/masking, i.e. it will remove
5804 any relevant symlinks both in /run and /etc.
5806 * Note that all long-running system services shipped with systemd will
5807 now default to a system call allow list (rather than a deny list, as
5808 before). In particular, systemd-udevd will now enforce one too. For
5809 most cases this should be safe, however downstream distributions
5810 which disabled sandboxing of systemd-udevd (specifically the
5811 MountFlags= setting), might want to disable this security feature
5812 too, as the default allow-listing will prohibit all mount, swap,
5813 reboot and clock changing operations from udev rules.
5815 * sd-boot acquired new loader configuration settings to optionally turn
5816 off Windows and MacOS boot partition discovery as well as
5817 reboot-into-firmware menu items. It is also able to pick a better
5818 screen resolution for HiDPI systems, and now provides loader
5819 configuration settings to change the resolution explicitly.
5821 * systemd-resolved now supports DNS-over-TLS. It's still
5822 turned off by default, use DNSOverTLS=opportunistic to turn it on in
5823 resolved.conf. We intend to make this the default as soon as couple
5824 of additional techniques for optimizing the initial latency caused by
5825 establishing a TLS/TCP connection are implemented.
5827 * systemd-resolved.service and systemd-networkd.service now set
5828 DynamicUser=yes. The users systemd-resolve and systemd-network are
5829 not created by systemd-sysusers anymore.
5831 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
5832 that embed a network facing module into any process using getpwuid()
5833 or related call: the dynamic allocation of the user ID for
5834 systemd-resolved.service means the service manager has to check NSS
5835 if the user name is already taken when forking off the service. Since
5836 the user in the common case won't be defined in /etc/passwd the
5837 lookup is likely to trigger nss-ldap which in turn might use NSS to
5838 ask systemd-resolved for hostname lookups. This will hence result in
5839 a deadlock: a user name lookup in order to start
5840 systemd-resolved.service will result in a hostname lookup for which
5841 systemd-resolved.service needs to be started already. There are
5842 multiple ways to work around this problem: pre-allocate the
5843 "systemd-resolve" user on such systems, so that nss-ldap won't be
5844 triggered; or use a different NSS package that doesn't do networking
5845 in-process but provides a local asynchronous name cache; or configure
5846 the NSS package to avoid lookups for UIDs in the range `pkg-config
5847 systemd --variable=dynamicuidmin` … `pkg-config systemd
5848 --variable=dynamicuidmax`, so that it does not consider itself
5849 authoritative for the same UID range systemd allocates dynamic users
5852 * The systemd-resolve tool has been renamed to resolvectl (it also
5853 remains available under the old name, for compatibility), and its
5854 interface is now verb-based, similar in style to the other <xyz>ctl
5855 tools, such as systemctl or loginctl.
5857 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
5858 compatibility. It may be symlinked under the 'resolvconf' name, in
5859 which case it will take arguments and input compatible with the
5860 Debian and FreeBSD resolvconf tool.
5862 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
5863 where the system initially suspends, and after a timeout resumes and
5866 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
5867 set the client will only send a DUID as client identifier.
5869 * The nss-systemd glibc NSS module will now enumerate dynamic users and
5870 groups in effect. Previously, it could resolve UIDs/GIDs to user
5871 names/groups and vice versa, but did not support enumeration.
5873 * journald's Compress= configuration setting now optionally accepts a
5874 byte threshold value. All journal objects larger than this threshold
5875 will be compressed, smaller ones will not. Previously this threshold
5876 was not configurable and set to 512.
5878 * A new system.conf setting NoNewPrivileges= is now available which may
5879 be used to turn off acquisition of new privileges system-wide
5880 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
5881 for all its children). Note that turning this option on means setuid
5882 binaries and file system capabilities lose their special powers.
5883 While turning on this option is a big step towards a more secure
5884 system, doing so is likely to break numerous pre-existing UNIX tools,
5885 in particular su and sudo.
5887 * A new service systemd-time-sync-wait.service has been added. If
5888 enabled it will delay the time-sync.target unit at boot until time
5889 synchronization has been received from the network. This
5890 functionality is useful on systems lacking a local RTC or where it is
5891 acceptable that the boot process shall be delayed by external network
5894 * When hibernating, systemd will now inform the kernel of the image
5895 write offset, on kernels new enough to support this. This means swap
5896 files should work for hibernation now.
5898 * When loading unit files, systemd will now look for drop-in unit files
5899 extensions in additional places. Previously, for a unit file name
5900 "foo-bar-baz.service" it would look for dropin files in
5901 "foo-bar-baz.service.d/*.conf". Now, it will also look in
5902 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
5903 service name truncated after all inner dashes. This scheme allows
5904 writing drop-ins easily that apply to a whole set of unit files at
5905 once. It's particularly useful for mount and slice units (as their
5906 naming is prefix based), but is also useful for service and other
5907 units, for packages that install multiple unit files at once,
5908 following a strict naming regime of beginning the unit file name with
5909 the package's name. Two new specifiers are now supported in unit
5910 files to match this: %j and %J are replaced by the part of the unit
5911 name following the last dash.
5913 * Unit files and other configuration files that support specifier
5914 expansion now understand another three new specifiers: %T and %V will
5915 resolve to /tmp and /var/tmp respectively, or whatever temporary
5916 directory has been set for the calling user. %E will expand to either
5917 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
5919 * The ExecStart= lines of unit files are no longer required to
5920 reference absolute paths. If non-absolute paths are specified the
5921 specified binary name is searched within the service manager's
5922 built-in $PATH, which may be queried with 'systemd-path
5923 search-binaries-default'. It's generally recommended to continue to
5924 use absolute paths for all binaries specified in unit files.
5926 * Units gained a new load state "bad-setting", which is used when a
5927 unit file was loaded, but contained fatal errors which prevent it
5928 from being started (for example, a service unit has been defined
5929 lacking both ExecStart= and ExecStop= lines).
5931 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
5932 support alternative debuggers, for example lldb. The old name
5933 continues to be available however, for compatibility reasons. Use the
5934 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
5935 to pick an alternative debugger instead of the default gdb.
5937 * systemctl and the other tools will now output escape sequences that
5938 generate proper clickable hyperlinks in various terminal emulators
5939 where useful (for example, in the "systemctl status" output you can
5940 now click on the unit file name to quickly open it in the
5941 editor/viewer of your choice). Note that not all terminal emulators
5942 support this functionality yet, but many do. Unfortunately, the
5943 "less" pager doesn't support this yet, hence this functionality is
5944 currently automatically turned off when a pager is started (which
5945 happens quite often due to auto-paging). We hope to remove this
5946 limitation as soon as "less" learns these escape sequences. This new
5947 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
5948 environment variable. For details on these escape sequences see:
5949 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
5951 * networkd's .network files now support a new IPv6MTUBytes= option for
5952 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
5953 option in the [Route] section to configure the MTU to use for
5954 specific routes. It also gained support for configuration of the DHCP
5955 "UserClass" option through the new UserClass= setting. It gained
5956 three new options in the new [CAN] section for configuring CAN
5957 networks. The MULTICAST and ALLMULTI interface flags may now be
5958 controlled explicitly with the new Multicast= and AllMulticast=
5961 * networkd will now automatically make use of the kernel's route
5962 expiration feature, if it is available.
5964 * udevd's .link files now support setting the number of receive and
5965 transmit channels, using the RxChannels=, TxChannels=,
5966 OtherChannels=, CombinedChannels= settings.
5968 * Support for UDPSegmentationOffload= has been removed, given its
5969 limited support in hardware, and waning software support.
5971 * networkd's .netdev files now support creating "netdevsim" interfaces.
5973 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
5974 to query the unit belonging to a specific kernel control group.
5976 * systemd-analyze gained a new verb "cat-config", which may be used to
5977 dump the contents of any configuration file, with all its matching
5978 drop-in files added in, and honouring the usual search and masking
5979 logic applied to systemd configuration files. For example use
5980 "systemd-analyze cat-config systemd/system.conf" to get the complete
5981 system configuration file of systemd how it would be loaded by PID 1
5982 itself. Similar to this, various tools such as systemd-tmpfiles or
5983 systemd-sysusers, gained a new option "--cat-config", which does the
5984 corresponding operation for their own configuration settings. For
5985 example, "systemd-tmpfiles --cat-config" will now output the full
5986 list of tmpfiles.d/ lines in place.
5988 * timedatectl gained three new verbs: "show" shows bus properties of
5989 systemd-timedated, "timesync-status" shows the current NTP
5990 synchronization state of systemd-timesyncd, and "show-timesync"
5991 shows bus properties of systemd-timesyncd.
5993 * systemd-timesyncd gained a bus interface on which it exposes details
5996 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
5997 understood by systemd-timedated. It takes a colon-separated list of
5998 unit names of NTP client services. The list is used by
5999 "timedatectl set-ntp".
6001 * systemd-nspawn gained a new --rlimit= switch for setting initial
6002 resource limits for the container payload. There's a new switch
6003 --hostname= to explicitly override the container's hostname. A new
6004 --no-new-privileges= switch may be used to control the
6005 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
6006 --oom-score-adjust= switch controls the OOM scoring adjustment value
6007 for the payload. The new --cpu-affinity= switch controls the CPU
6008 affinity of the container payload. The new --resolv-conf= switch
6009 allows more detailed control of /etc/resolv.conf handling of the
6010 container. Similarly, the new --timezone= switch allows more detailed
6011 control of /etc/localtime handling of the container.
6013 * systemd-detect-virt gained a new --list switch, which will print a
6014 list of all currently known VM and container environments.
6016 * Support for "Portable Services" has been added, see
6017 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
6018 experimental, but this is expected to change soon. Reflecting this
6019 experimental state, the "portablectl" binary is not installed into
6020 /usr/bin yet. The binary has to be called with the full path
6021 /usr/lib/systemd/portablectl instead.
6023 * journalctl's and systemctl's -o switch now knows a new log output
6024 mode "with-unit". The output it generates is very similar to the
6025 regular "short" mode, but displays the unit name instead of the
6026 syslog tag for each log line. Also, the date is shown with timezone
6027 information. This mode is probably more useful than the classic
6028 "short" output mode for most purposes, except where pixel-perfect
6029 compatibility with classic /var/log/messages formatting is required.
6031 * A new --dump-bus-properties switch has been added to the systemd
6032 binary, which may be used to dump all supported D-Bus properties.
6033 (Options which are still supported, but are deprecated, are *not*
6036 * sd-bus gained a set of new calls:
6037 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
6038 enable/disable the "floating" state of a bus slot object,
6039 i.e. whether the slot object pins the bus it is allocated for into
6040 memory or if the bus slot object gets disconnected when the bus goes
6041 away. sd_bus_open_with_description(),
6042 sd_bus_open_user_with_description(),
6043 sd_bus_open_system_with_description() may be used to allocate bus
6044 objects and set their description string already during allocation.
6046 * sd-event gained support for watching inotify events from the event
6047 loop, in an efficient way, sharing inotify handles between multiple
6048 users. For this a new function sd_event_add_inotify() has been added.
6050 * sd-event and sd-bus gained support for calling special user-supplied
6051 destructor functions for userdata pointers associated with
6052 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
6053 functions sd_bus_slot_set_destroy_callback,
6054 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
6055 sd_bus_track_get_destroy_callback,
6056 sd_event_source_set_destroy_callback,
6057 sd_event_source_get_destroy_callback have been added.
6059 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
6061 * PID 1 will now automatically reschedule .timer units whenever the
6062 local timezone changes. (They previously got rescheduled
6063 automatically when the system clock changed.)
6065 * New documentation has been added to document cgroups delegation,
6066 portable services and the various code quality tools we have set up:
6068 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
6069 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
6070 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
6072 * The Boot Loader Specification has been added to the source tree.
6074 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
6076 While moving it into our source tree we have updated it and further
6077 changes are now accepted through the usual github PR workflow.
6079 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
6080 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
6081 earlier PAM modules. The data in these fields is used to initialize
6082 the session scope's resource properties. Thus external PAM modules
6083 may now configure per-session limits, for example sourced from
6084 external user databases.
6086 * socket units with Accept=yes will now maintain a "refused" counter in
6087 addition to the existing "accepted" counter, counting connections
6088 refused due to the enforced limits.
6090 * The "systemd-path search-binaries-default" command may now be use to
6091 query the default, built-in $PATH PID 1 will pass to the services it
6094 * A new unit file setting PrivateMounts= has been added. It's a boolean
6095 option. If enabled the unit's processes are invoked in their own file
6096 system namespace. Note that this behaviour is also implied if any
6097 other file system namespacing options (such as PrivateTmp=,
6098 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
6099 primarily useful for services that do not use any of the other file
6100 system namespacing options. One such service is systemd-udevd.service
6101 where this is now used by default.
6103 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
6104 when the system is booted in UEFI "secure mode".
6106 * A new unit "system-update-pre.target" is added, which defines an
6107 optional synchronization point for offline system updates, as
6108 implemented by the pre-existing "system-update.target" unit. It
6109 allows ordering services before the service that executes the actual
6110 update process in a generic way.
6112 * Systemd now emits warnings whenever .include syntax is used.
6114 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
6115 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
6116 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
6117 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
6118 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
6119 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
6120 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
6121 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
6122 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
6123 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
6124 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
6125 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
6126 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
6127 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
6128 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
6129 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
6130 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
6131 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
6132 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
6133 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
6134 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
6135 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
6136 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
6137 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
6138 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
6139 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
6140 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
6141 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
6142 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
6144 — Berlin, 2018-06-22
6148 * The MemoryAccounting= unit property now defaults to on. After
6149 discussions with the upstream control group maintainers we learnt
6150 that the negative impact of cgroup memory accounting on current
6151 kernels is finally relatively minimal, so that it should be safe to
6152 enable this by default without affecting system performance. Besides
6153 memory accounting only task accounting is turned on by default, all
6154 other forms of resource accounting (CPU, IO, IP) remain off for now,
6155 because it's not clear yet that their impact is small enough to move
6156 from opt-in to opt-out. We recommend downstreams to leave memory
6157 accounting on by default if kernel 4.14 or higher is primarily
6158 used. On very resource constrained systems or when support for old
6159 kernels is a necessity, -Dmemory-accounting-default=false can be used
6160 to revert this change.
6162 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
6163 %udev_rules_update) and the journal catalog (%journal_catalog_update)
6164 from the upgrade scriptlets of individual packages now do nothing.
6165 Transfiletriggers have been added which will perform those updates
6166 once at the end of the transaction.
6168 Similar transfiletriggers have been added to execute any sysctl.d
6169 and binfmt.d rules. Thus, it should be unnecessary to provide any
6170 scriptlets to execute this configuration from package installation
6173 * systemd-sysusers gained a mode where the configuration to execute is
6174 specified on the command line, but this configuration is not executed
6175 directly, but instead it is merged with the configuration on disk,
6176 and the result is executed. This is useful for package installation
6177 scripts which want to create the user before installing any files on
6178 disk (in case some of those files are owned by that user), while
6179 still allowing local admin overrides.
6181 This functionality is exposed to rpm scriptlets through a new
6182 %sysusers_create_package macro. Old %sysusers_create and
6183 %sysusers_create_inline macros are deprecated.
6185 A transfiletrigger for sysusers.d configuration is now installed,
6186 which means that it should be unnecessary to call systemd-sysusers from
6187 package installation scripts, unless the package installs any files
6188 owned by those newly-created users, in which case
6189 %sysusers_create_package should be used.
6191 * Analogous change has been done for systemd-tmpfiles: it gained a mode
6192 where the command-line configuration is merged with the configuration
6193 on disk. This is exposed as the new %tmpfiles_create_package macro,
6194 and %tmpfiles_create is deprecated. A transfiletrigger is installed
6195 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
6196 from package installation scripts.
6198 * sysusers.d configuration for a user may now also specify the group
6199 number, in addition to the user number ("u username 123:456"), or
6200 without the user number ("u username -:456").
6202 * Configution items for systemd-sysusers can now be specified as
6203 positional arguments when the new --inline switch is used.
6205 * The login shell of users created through sysusers.d may now be
6206 specified (previously, it was always /bin/sh for root and
6207 /sbin/nologin for other users).
6209 * systemd-analyze gained a new --global switch to look at global user
6210 configuration. It also gained a unit-paths verb to list the unit load
6211 paths that are compiled into systemd (which can be used with
6212 --systemd, --user, or --global).
6214 * udevadm trigger gained a new --settle/-w option to wait for any
6215 triggered events to finish (but just those, and not any other events
6216 which are triggered meanwhile).
6218 * The action that systemd-logind takes when the lid is closed and the
6219 machine is connected to external power can now be configured using
6220 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
6221 was determined by HandleLidSwitch=, and, for backwards compatibility,
6222 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
6224 * journalctl will periodically call sd_journal_process() to make it
6225 resilient against inotify queue overruns when journal files are
6226 rotated very quickly.
6228 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
6229 sd_bus_get_n_queued_write — may be used to check the number of
6230 pending bus messages.
6232 * systemd gained a new
6233 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
6234 which can be used to migrate foreign processes to scope and service
6235 units. The primary user for this new API is systemd itself: the
6236 systemd --user instance uses this call of the systemd --system
6237 instance to migrate processes if it itself gets the request to
6238 migrate processes and the kernel refuses this due to access
6239 restrictions. Thanks to this "systemd-run --scope --user …" works
6240 again in pure cgroup v2 environments when invoked from the user
6243 * A new TemporaryFileSystem= setting can be used to mask out part of
6244 the real file system tree with tmpfs mounts. This may be combined
6245 with BindPaths= and BindReadOnlyPaths= to hide files or directories
6246 not relevant to the unit, while still allowing some paths lower in
6247 the tree to be accessed.
6249 ProtectHome=tmpfs may now be used to hide user home and runtime
6250 directories from units, in a way that is mostly equivalent to
6251 "TemporaryFileSystem=/home /run/user /root".
6253 * Non-service units are now started with KeyringMode=shared by default.
6254 This means that mount and swapon and other mount tools have access
6255 to keys in the main keyring.
6257 * /sys/fs/bpf is now mounted automatically.
6259 * QNX virtualization is now detected by systemd-detect-virt and may
6260 be used in ConditionVirtualization=.
6262 * IPAccounting= may now be enabled also for slice units.
6264 * A new -Dsplit-bin= build configuration switch may be used to specify
6265 whether bin and sbin directories are merged, or if they should be
6266 included separately in $PATH and various listings of executable
6267 directories. The build configuration scripts will try to autodetect
6268 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
6269 system, but distributions are encouraged to configure this
6272 * A new -Dok-color= build configuration switch may be used to change
6273 the colour of "OK" status messages.
6275 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
6276 PrivateNetwork=yes was buggy in previous versions of systemd. This
6277 means that after the upgrade and daemon-reexec, any such units must
6280 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
6281 will not exclude read-only files owned by root from cleanup.
6283 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
6284 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
6285 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
6286 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
6287 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
6288 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
6289 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
6290 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
6291 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
6292 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
6293 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
6294 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
6295 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
6296 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
6297 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
6298 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
6300 — Warsaw, 2018-03-05
6304 * Some keyboards come with a zoom see-saw or rocker which until now got
6305 mapped to the Linux "zoomin/out" keys in hwdb. However, these
6306 keycodes are not recognized by any major desktop. They now produce
6307 Up/Down key events so that they can be used for scrolling.
6309 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
6310 slightly: previously, if an argument was specified for lines of this
6311 type (i.e. the right-most column was set) this string was appended to
6312 existing files each time systemd-tmpfiles was run. This behaviour was
6313 different from what the documentation said, and not particularly
6314 useful, as repeated systemd-tmpfiles invocations would not be
6315 idempotent and grow such files without bounds. With this release
6316 behaviour has been altered to match what the documentation says:
6317 lines of this type only have an effect if the indicated files don't
6318 exist yet, and only then the argument string is written to the file.
6320 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
6321 systemd-tmpfiles behaviour: previously, read-only files owned by root
6322 were always excluded from the file "aging" algorithm (i.e. the
6323 automatic clean-up of directories like /tmp based on
6324 atime/mtime/ctime). We intend to drop this restriction, and age files
6325 by default even when owned by root and read-only. This behaviour was
6326 inherited from older tools, but there have been requests to remove
6327 it, and it's not obvious why this restriction was made in the first
6328 place. Please speak up now, if you are aware of software that requires
6329 this behaviour, otherwise we'll remove the restriction in v238.
6331 * A new environment variable $SYSTEMD_OFFLINE is now understood by
6332 systemctl. It takes a boolean argument. If on, systemctl assumes it
6333 operates on an "offline" OS tree, and will not attempt to talk to the
6334 service manager. Previously, this mode was implicitly enabled if a
6335 chroot() environment was detected, and this new environment variable
6336 now provides explicit control.
6338 * .path and .socket units may now be created transiently, too.
6339 Previously only service, mount, automount and timer units were
6340 supported as transient units. The systemd-run tool has been updated
6341 to expose this new functionality, you may hence use it now to bind
6342 arbitrary commands to path or socket activation on-the-fly from the
6343 command line. Moreover, almost all properties are now exposed for the
6344 unit types that already supported transient operation.
6346 * The systemd-mount command gained support for a new --owner= parameter
6347 which takes a user name, which is then resolved and included in uid=
6348 and gid= mount options string of the file system to mount.
6350 * A new unit condition ConditionControlGroupController= has been added
6351 that checks whether a specific cgroup controller is available.
6353 * Unit files, udev's .link files, and systemd-networkd's .netdev and
6354 .network files all gained support for a new condition
6355 ConditionKernelVersion= for checking against specific kernel
6358 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
6359 support for configuring device flags in the Flags= setting. In the
6360 same files, the [Tunnel] section gained support for configuring
6361 AllowLocalRemote=. The [Route] section in .network files gained
6362 support for configuring InitialCongestionWindow=,
6363 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
6364 understands RapidCommit=.
6366 * systemd-networkd's DHCPv6 support gained support for Prefix
6369 * sd-bus gained support for a new "watch-bind" feature. When this
6370 feature is enabled, an sd_bus connection may be set up to connect to
6371 an AF_UNIX socket in the file system as soon as it is created. This
6372 functionality is useful for writing early-boot services that
6373 automatically connect to the system bus as soon as it is started,
6374 without ugly time-based polling. systemd-networkd and
6375 systemd-resolved have been updated to make use of this
6376 functionality. busctl exposes this functionality in a new
6377 --watch-bind= command line switch.
6379 * sd-bus will now optionally synthesize a local "Connected" signal as
6380 soon as a D-Bus connection is set up fully. This message mirrors the
6381 already existing "Disconnected" signal which is synthesized when the
6382 connection is terminated. This signal is generally useful but
6383 particularly handy in combination with the "watch-bind" feature
6384 described above. Synthesizing of this message has to be requested
6385 explicitly through the new API call sd_bus_set_connected_signal(). In
6386 addition a new call sd_bus_is_ready() has been added that checks
6387 whether a connection is fully set up (i.e. between the "Connected" and
6388 "Disconnected" signals).
6390 * sd-bus gained two new calls sd_bus_request_name_async() and
6391 sd_bus_release_name_async() for asynchronously registering bus
6392 names. Similar, there is now sd_bus_add_match_async() for installing
6393 a signal match asynchronously. All of systemd's own services have
6394 been updated to make use of these calls. Doing these operations
6395 asynchronously has two benefits: it reduces the risk of deadlocks in
6396 case of cyclic dependencies between bus services, and it speeds up
6397 service initialization since synchronization points for bus
6398 round-trips are removed.
6400 * sd-bus gained two new calls sd_bus_match_signal() and
6401 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
6402 and sd_bus_add_match_async() but instead of taking a D-Bus match
6403 string take match fields as normal function parameters.
6405 * sd-bus gained two new calls sd_bus_set_sender() and
6406 sd_bus_message_set_sender() for setting the sender name of outgoing
6407 messages (either for all outgoing messages or for just one specific
6408 one). These calls are only useful in direct connections as on
6409 brokered connections the broker fills in the sender anyway,
6410 overwriting whatever the client filled in.
6412 * sd-event gained a new pseudo-handle that may be specified on all API
6413 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
6414 used this refers to the default event loop object of the calling
6415 thread. Note however that this does not implicitly allocate one —
6416 which has to be done prior by using sd_event_default(). Similarly
6417 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
6418 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
6419 to the default bus of the specified type of the calling thread. Here
6420 too this does not implicitly allocate bus connection objects, this
6421 has to be done prior with sd_bus_default() and friends.
6423 * sd-event gained a new call pair
6424 sd_event_source_{get|set}_io_fd_own(). This may be used to request
6425 automatic closure of the file descriptor an IO event source watches
6426 when the event source is destroyed.
6428 * systemd-networkd gained support for natively configuring WireGuard
6431 * In previous versions systemd synthesized user records both for the
6432 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
6433 internally. In order to simplify distribution-wide renames of the
6434 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
6435 new transitional flag file has been added: if
6436 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
6437 user and group record within the systemd codebase is disabled.
6439 * systemd-notify gained a new --uid= option for selecting the source
6440 user/UID to use for notification messages sent to the service
6443 * journalctl gained a new --grep= option to list only entries in which
6444 the message matches a certain pattern. By default matching is case
6445 insensitive if the pattern is lowercase, and case sensitive
6446 otherwise. Option --case-sensitive=yes|no can be used to override
6447 this an specify case sensitivity or case insensitivity.
6449 * There's now a "systemd-analyze service-watchdogs" command for printing
6450 the current state of the service runtime watchdog, and optionally
6451 enabling or disabling the per-service watchdogs system-wide if given a
6452 boolean argument (i.e. the concept you configure in WatchdogSec=), for
6453 debugging purposes. There's also a kernel command line option
6454 systemd.service_watchdogs= for controlling the same.
6456 * Two new "log-level" and "log-target" options for systemd-analyze were
6457 added that merge the now deprecated get-log-level, set-log-level and
6458 get-log-target, set-log-target pairs. The deprecated options are still
6459 understood for backwards compatibility. The two new options print the
6460 current value when no arguments are given, and set them when a
6461 level/target is given as an argument.
6463 * sysusers.d's "u" lines now optionally accept both a UID and a GID
6464 specification, separated by a ":" character, in order to create users
6465 where UID and GID do not match.
6467 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
6468 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
6469 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
6470 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
6471 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
6472 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
6473 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
6474 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
6475 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
6476 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
6477 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
6478 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
6479 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
6480 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
6481 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
6482 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
6483 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
6484 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
6485 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
6486 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
6493 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
6494 in v235 has been extended to also set the dummy.ko module option
6495 numdummies=0, preventing the kernel from automatically creating
6496 dummy0. All dummy interfaces must now be explicitly created.
6498 * Unknown '%' specifiers in configuration files are now rejected. This
6499 applies to units and tmpfiles.d configuration. Any percent characters
6500 that are followed by a letter or digit that are not supposed to be
6501 interpreted as the beginning of a specifier should be escaped by
6502 doubling ("%%"). (So "size=5%" is still accepted, as well as
6503 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
6504 valid specifiers today.)
6506 * systemd-resolved now maintains a new dynamic
6507 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
6508 recommended to make /etc/resolv.conf a symlink to it. This file
6509 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
6510 includes dynamically acquired search domains, achieving more correct
6511 DNS resolution by software that bypasses local DNS APIs such as NSS.
6513 * The "uaccess" udev tag has been dropped from /dev/kvm and
6514 /dev/dri/renderD*. These devices now have the 0666 permissions by
6515 default (but this may be changed at build-time). /dev/dri/renderD*
6516 will now be owned by the "render" group along with /dev/kfd.
6518 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
6519 systemd-journal-gatewayd.service and
6520 systemd-journal-upload.service. This means "nss-systemd" must be
6521 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
6522 services are resolved properly.
6524 * In /etc/fstab two new mount options are now understood:
6525 x-systemd.makefs and x-systemd.growfs. The former has the effect that
6526 the configured file system is formatted before it is mounted, the
6527 latter that the file system is resized to the full block device size
6528 after it is mounted (i.e. if the file system is smaller than the
6529 partition it resides on, it's grown). This is similar to the fsck
6530 logic in /etc/fstab, and pulls in systemd-makefs@.service and
6531 systemd-growfs@.service as necessary, similar to
6532 systemd-fsck@.service. Resizing is currently only supported on ext4
6535 * In systemd-networkd, the IPv6 RA logic now optionally may announce
6536 DNS server and domain information.
6538 * Support for the LUKS2 on-disk format for encrypted partitions has
6539 been added. This requires libcryptsetup2 during compilation and
6542 * The systemd --user instance will now signal "readiness" when its
6543 basic.target unit has been reached, instead of when the run queue ran
6544 empty for the first time.
6546 * Tmpfiles.d with user configuration are now also supported.
6547 systemd-tmpfiles gained a new --user switch, and snippets placed in
6548 ~/.config/user-tmpfiles.d/ and corresponding directories will be
6549 executed by systemd-tmpfiles --user running in the new
6550 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
6551 running in the user session.
6553 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
6554 %S resolves to the top-level state directory (/var/lib for the system
6555 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
6556 top-level cache directory (/var/cache for the system instance,
6557 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
6558 logs directory (/var/log for the system instance,
6559 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
6560 existing %t specifier, that resolves to the top-level runtime
6561 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
6564 * journalctl learnt a new parameter --output-fields= for limiting the
6565 set of journal fields to output in verbose and JSON output modes.
6567 * systemd-timesyncd's configuration file gained a new option
6568 RootDistanceMaxSec= for setting the maximum root distance of servers
6569 it'll use, as well as the new options PollIntervalMinSec= and
6570 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
6572 * bootctl gained a new command "list" for listing all available boot
6573 menu items on systems that follow the boot loader specification.
6575 * systemctl gained a new --dry-run switch that shows what would be done
6576 instead of doing it, and is currently supported by the shutdown and
6579 * ConditionSecurity= can now detect the TOMOYO security module.
6581 * Unit file [Install] sections are now also respected in unit drop-in
6582 files. This is intended to be used by drop-ins under /usr/lib/.
6584 * systemd-firstboot may now also set the initial keyboard mapping.
6586 * Udev "changed" events for devices which are exposed as systemd
6587 .device units are now propagated to units specified in
6588 ReloadPropagatedFrom= as reload requests.
6590 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
6591 unit template name (i.e. a name in the form of 'foobar@.service',
6592 without the instance component between the '@' and - the '.'), then
6593 the escaped sysfs path of the device is automatically used as the
6596 * SystemCallFilter= in unit files has been extended so that an "errno"
6597 can be specified individually for each system call. Example:
6598 SystemCallFilter=~uname:EILSEQ.
6600 * The cgroup delegation logic has been substantially updated. Delegate=
6601 now optionally takes a list of controllers (instead of a boolean, as
6602 before), which lists the controllers to delegate at least.
6604 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
6606 * A new LogLevelMax= setting configures the maximum log level any
6607 process of the service may log at (i.e. anything with a lesser
6608 priority than what is specified is automatically dropped). A new
6609 LogExtraFields= setting allows configuration of additional journal
6610 fields to attach to all log records generated by any of the unit's
6613 * New StandardInputData= and StandardInputText= settings along with the
6614 new option StandardInput=data may be used to configure textual or
6615 binary data that shall be passed to the executed service process via
6616 standard input, encoded in-line in the unit file.
6618 * StandardInput=, StandardOutput= and StandardError= may now be used to
6619 connect stdin/stdout/stderr of executed processes directly with a
6620 file or AF_UNIX socket in the file system, using the new "file:" option.
6622 * A new unit file option CollectMode= has been added, that allows
6623 tweaking the garbage collection logic for units. It may be used to
6624 tell systemd to garbage collect units that have failed automatically
6625 (normally it only GCs units that exited successfully). systemd-run
6626 and systemd-mount expose this new functionality with a new -G option.
6628 * "machinectl bind" may now be used to bind mount non-directories
6629 (i.e. regularfiles, devices, fifos, sockets).
6631 * systemd-analyze gained a new verb "calendar" for validating and
6632 testing calendar time specifications to use for OnCalendar= in timer
6633 units. Besides validating the expression it will calculate the next
6634 time the specified expression would elapse.
6636 * In addition to the pre-existing FailureAction= unit file setting
6637 there's now SuccessAction=, for configuring a shutdown action to
6638 execute when a unit completes successfully. This is useful in
6639 particular inside containers that shall terminate after some workload
6640 has been completed. Also, both options are now supported for all unit
6641 types, not just services.
6643 * networkds's IP rule support gained two new options
6644 IncomingInterface= and OutgoingInterface= for configuring the incoming
6645 and outgoing interfaces of configured rules. systemd-networkd also
6646 gained support for "vxcan" network devices.
6648 * networkd gained a new setting RequiredForOnline=, taking a
6649 boolean. If set, systemd-wait-online will take it into consideration
6650 when determining that the system is up, otherwise it will ignore the
6651 interface for this purpose.
6653 * The sd_notify() protocol gained support for a new operation: with
6654 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
6655 store again, ahead of POLLHUP or POLLERR when they are removed
6658 * A new document doc/UIDS-GIDS.md has been added to the source tree,
6659 that documents the UID/GID range and assignment assumptions and
6660 requirements of systemd.
6662 * The watchdog device PID 1 will ping may now be configured through the
6663 WatchdogDevice= configuration file setting, or by setting the
6664 systemd.watchdog_service= kernel commandline option.
6666 * systemd-resolved's gained support for registering DNS-SD services on
6667 the local network using MulticastDNS. Services may either be
6668 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
6669 the same dir below /run, /usr/lib), or through its D-Bus API.
6671 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
6672 extend the effective start, runtime, and stop time. The service must
6673 continue to send EXTEND_TIMEOUT_USEC within the period specified to
6674 prevent the service manager from making the service as timedout.
6676 * systemd-resolved's DNSSEC support gained support for RFC 8080
6677 (Ed25519 keys and signatures).
6679 * The systemd-resolve command line tool gained a new set of options
6680 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
6681 --set-nta= and --revert to configure per-interface DNS configuration
6682 dynamically during runtime. It's useful for pushing DNS information
6683 into systemd-resolved from DNS hook scripts that various interface
6684 managing software supports (such as pppd).
6686 * systemd-nspawn gained a new --network-namespace-path= command line
6687 option, which may be used to make a container join an existing
6688 network namespace, by specifying a path to a "netns" file.
6690 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
6691 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
6692 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
6693 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
6694 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
6695 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
6696 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
6697 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
6698 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
6699 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
6700 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
6701 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
6702 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
6703 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
6704 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
6705 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
6706 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
6707 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
6708 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
6709 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
6710 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
6711 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
6712 Jędrzejewski-Szmek, Zeal Jagannatha
6714 — Berlin, 2017-12-14
6718 * INCOMPATIBILITY: systemd-logind.service and other long-running
6719 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
6720 communication with the outside. This generally improves security of
6721 the system, and is in almost all cases a safe and good choice, as
6722 these services do not and should not provide any network-facing
6723 functionality. However, systemd-logind uses the glibc NSS API to
6724 query the user database. This creates problems on systems where NSS
6725 is set up to directly consult network services for user database
6726 lookups. In particular, this creates incompatibilities with the
6727 "nss-nis" module, which attempts to directly contact the NIS/YP
6728 network servers it is configured for, and will now consistently
6729 fail. In such cases, it is possible to turn off IP sandboxing for
6730 systemd-logind.service (set IPAddressDeny= in its [Service] section
6731 to the empty string, via a .d/ unit file drop-in). Downstream
6732 distributions might want to update their nss-nis packaging to include
6733 such a drop-in snippet, accordingly, to hide this incompatibility
6734 from the user. Another option is to make use of glibc's nscd service
6735 to proxy such network requests through a privilege-separated, minimal
6736 local caching daemon, or to switch to more modern technologies such
6737 sssd, whose NSS hook-ups generally do not involve direct network
6738 access. In general, we think it's definitely time to question the
6739 implementation choices of nss-nis, i.e. whether it's a good idea
6740 today to embed a network-facing loadable module into all local
6741 processes that need to query the user database, including the most
6742 trivial and benign ones, such as "ls". For more details about
6743 IPAddressDeny= see below.
6745 * A new modprobe.d drop-in is now shipped by default that sets the
6746 bonding module option max_bonds=0. This overrides the kernel default,
6747 to avoid conflicts and ambiguity as to whether or not bond0 should be
6748 managed by systemd-networkd or not. This resolves multiple issues
6749 with bond0 properties not being applied, when bond0 is configured
6750 with systemd-networkd. Distributors may choose to not package this,
6751 however in that case users will be prevented from correctly managing
6752 bond0 interface using systemd-networkd.
6754 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
6755 which print the logging level and target of the system manager. They
6756 complement the existing "set-log-level" and "set-log-target" verbs
6757 used to change those values.
6759 * journald.conf gained a new boolean setting ReadKMsg= which defaults
6760 to on. If turned off kernel log messages will not be read by
6761 systemd-journald or included in the logs. It also gained a new
6762 setting LineMax= for configuring the maximum line length in
6763 STDOUT/STDERR log streams. The new default for this value is 48K, up
6764 from the previous hardcoded 2048.
6766 * A new unit setting RuntimeDirectoryPreserve= has been added, which
6767 allows more detailed control of what to do with a runtime directory
6768 configured with RuntimeDirectory= (i.e. a directory below /run or
6769 $XDG_RUNTIME_DIR) after a unit is stopped.
6771 * The RuntimeDirectory= setting for units gained support for creating
6772 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
6773 one top-level directory.
6775 * Units gained new options StateDirectory=, CacheDirectory=,
6776 LogsDirectory= and ConfigurationDirectory= which are closely related
6777 to RuntimeDirectory= but manage per-service directories below
6778 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
6779 possible to write unit files which when activated automatically gain
6780 properly owned service specific directories in these locations, thus
6781 making unit files self-contained and increasing compatibility with
6782 stateless systems and factory reset where /etc or /var are
6783 unpopulated at boot. Matching these new settings there's also
6784 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
6785 ConfigurationDirectoryMode= for configuring the access mode of these
6786 directories. These settings are particularly useful in combination
6787 with DynamicUser=yes as they provide secure, properly-owned,
6788 writable, and stateful locations for storage, excluded from the
6789 sandbox that such services live in otherwise.
6791 * Automake support has been removed from this release. systemd is now
6794 * systemd-journald will now aggressively cache client metadata during
6795 runtime, speeding up log write performance under pressure. This comes
6796 at a small price though: as much of the metadata is read
6797 asynchronously from /proc/ (and isn't implicitly attached to log
6798 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
6799 metadata stored alongside a log entry might be slightly
6800 out-of-date. Previously it could only be slightly newer than the log
6801 message. The time window is small however, and given that the kernel
6802 is unlikely to be improved anytime soon in this regard, this appears
6805 * nss-myhostname/systemd-resolved will now by default synthesize an
6806 A/AAAA resource record for the "_gateway" hostname, pointing to the
6807 current default IP gateway. Previously it did that for the "gateway"
6808 name, hampering adoption, as some distributions wanted to leave that
6809 hostname open for local use. The old behaviour may still be
6810 requested at build time.
6812 * systemd-networkd's [Address] section in .network files gained a new
6813 Scope= setting for configuring the IP address scope. The [Network]
6814 section gained a new boolean setting ConfigureWithoutCarrier= that
6815 tells systemd-networkd to ignore link sensing when configuring the
6816 device. The [DHCP] section gained a new Anonymize= boolean option for
6817 turning on a number of options suggested in RFC 7844. A new
6818 [RoutingPolicyRule] section has been added for configuring the IP
6819 routing policy. The [Route] section has gained support for a new
6820 Type= setting which permits configuring
6821 blackhole/unreachable/prohibit routes.
6823 * The [VRF] section in .netdev files gained a new Table= setting for
6824 configuring the routing table to use. The [Tunnel] section gained a
6825 new Independent= boolean field for configuring tunnels independent of
6826 an underlying network interface. The [Bridge] section gained a new
6827 GroupForwardMask= option for configuration of propagation of link
6828 local frames between bridge ports.
6830 * The WakeOnLan= setting in .link files gained support for a number of
6831 new modes. A new TCP6SegmentationOffload= setting has been added for
6832 configuring TCP/IPv6 hardware segmentation offload.
6834 * The IPv6 RA sender implementation may now optionally send out RDNSS
6835 and RDNSSL records to supply DNS configuration to peers.
6837 * systemd-nspawn gained support for a new --system-call-filter= command
6838 line option for adding and removing entries in the default system
6839 call filter it applies. Moreover systemd-nspawn has been changed to
6840 implement a system call allow list instead of a deny list.
6842 * systemd-run gained support for a new --pipe command line option. If
6843 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
6844 are directly passed on to the activated transient service
6845 executable. This allows invoking arbitrary processes as systemd
6846 services (for example to take benefit of dependency management,
6847 accounting management, resource management or log management that is
6848 done automatically for services) — while still allowing them to be
6849 integrated in a classic UNIX shell pipeline.
6851 * When a service sends RELOAD=1 via sd_notify() and reload propagation
6852 using ReloadPropagationTo= is configured, a reload is now propagated
6853 to configured units. (Previously this was only done on explicitly
6854 requested reloads, using "systemctl reload" or an equivalent
6857 * For each service unit a restart counter is now kept: it is increased
6858 each time the service is restarted due to Restart=, and may be
6859 queried using "systemctl show -p NRestarts …".
6861 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
6862 @signal and @timer have been added, for usage with SystemCallFilter=
6863 in unit files and the new --system-call-filter= command line option
6864 of systemd-nspawn (see above).
6866 * ExecStart= lines in unit files gained two new modifiers: when a
6867 command line is prefixed with "!" the command will be executed as
6868 configured, except for the credentials applied by
6869 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
6870 "+", but does still apply namespacing options unlike "+". There's
6871 also "!!" now, which is mostly identical, but becomes a NOP on
6872 systems that support ambient capabilities. This is useful to write
6873 unit files that work with ambient capabilities where possible but
6874 automatically fall back to traditional privilege dropping mechanisms
6875 on systems where this is not supported.
6877 * ListenNetlink= settings in socket units now support RDMA netlink
6880 * A new unit file setting LockPersonality= has been added which permits
6881 locking down the chosen execution domain ("personality") of a service
6884 * A new special target "getty-pre.target" has been added, which is
6885 ordered before all text logins, and may be used to order services
6886 before textual logins acquire access to the console.
6888 * systemd will now attempt to load the virtio-rng.ko kernel module very
6889 early on if a VM environment supporting this is detected. This should
6890 improve entropy during early boot in virtualized environments.
6892 * A _netdev option is now supported in /etc/crypttab that operates in a
6893 similar way as the same option in /etc/fstab: it permits configuring
6894 encrypted devices that need to be ordered after the network is up.
6895 Following this logic, two new special targets
6896 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
6897 added that are to cryptsetup.target what remote-fs.target and
6898 remote-fs-pre.target are to local-fs.target.
6900 * Service units gained a new UnsetEnvironment= setting which permits
6901 unsetting specific environment variables for services that are
6902 normally passed to it (for example in order to mask out locale
6903 settings for specific services that can't deal with it).
6905 * Units acquired a new boolean option IPAccounting=. When turned on, IP
6906 traffic accounting (packet count as well as byte count) is done for
6907 the service, and shown as part of "systemctl status" or "systemd-run
6910 * Service units acquired two new options IPAddressAllow= and
6911 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
6912 for configuring a simple IP access control list for all sockets of
6913 the unit. These options are available also on .slice and .socket
6914 units, permitting flexible access list configuration for individual
6915 services as well as groups of services (as defined by a slice unit),
6916 including system-wide. Note that IP ACLs configured this way are
6917 enforced on every single IPv4 and IPv6 socket created by any process
6918 of the service unit, and apply to ingress as well as egress traffic.
6920 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
6921 structured log message is generated each time the unit is stopped,
6922 containing information about the consumed resources of this
6925 * A new setting KeyringMode= has been added to unit files, which may be
6926 used to control how the kernel keyring is set up for executed
6929 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
6930 "systemctl kexec" and "systemctl exit" are now always asynchronous in
6931 behaviour (that is: these commands return immediately after the
6932 operation was enqueued instead of waiting for the operation to
6933 complete). Previously, "systemctl poweroff" and "systemctl reboot"
6934 were asynchronous on systems using systemd-logind (i.e. almost
6935 always, and like they were on sysvinit), and the other three commands
6936 were unconditionally synchronous. With this release this is cleaned
6937 up, and callers will see the same asynchronous behaviour on all
6938 systems for all five operations.
6940 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
6943 * .timer units now accept calendar specifications in other timezones
6944 than UTC or the local timezone.
6946 * The tmpfiles snippet var.conf has been changed to create
6947 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
6948 the "utmp" group already, and it appears to be generally understood
6949 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
6950 databases. Previously this was implemented correctly for all these
6951 databases excepts btmp, which has been opened up like this now
6952 too. Note that while the other databases are world-readable
6953 (i.e. 0644), btmp is not and remains more restrictive.
6955 * The systemd-resolve tool gained a new --reset-server-features
6956 switch. When invoked like this systemd-resolved will forget
6957 everything it learnt about the features supported by the configured
6958 upstream DNS servers, and restarts the feature probing logic on the
6959 next resolver look-up for them at the highest feature level
6962 * The status dump systemd-resolved sends to the logs upon receiving
6963 SIGUSR1 now also includes information about all DNS servers it is
6964 configured to use, and the features levels it probed for them.
6966 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
6967 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
6968 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
6969 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
6970 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
6971 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
6972 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
6973 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
6974 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
6975 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
6976 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
6977 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
6978 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
6979 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
6980 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
6981 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
6982 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
6983 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
6984 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
6985 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
6987 — Berlin, 2017-10-06
6991 * Meson is now supported as build system in addition to Automake. It is
6992 our plan to remove Automake in one of our next releases, so that
6993 Meson becomes our exclusive build system. Hence, please start using
6994 the Meson build system in your downstream packaging. There's plenty
6995 of documentation around how to use Meson, the extremely brief
6998 ./autogen.sh && ./configure && make && sudo make install
7002 meson build && ninja -C build && sudo ninja -C build install
7004 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
7005 which permits configuring a timeout on the time a job is
7006 running. This is particularly useful for setting timeouts on jobs for
7009 * Unit files gained two new options ConditionUser= and ConditionGroup=
7010 for conditionalizing units based on the identity of the user/group
7011 running a systemd user instance.
7013 * systemd-networkd now understands a new FlowLabel= setting in the
7014 [VXLAN] section of .network files, as well as a Priority= in
7015 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
7016 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
7017 gained support for configuration of GENEVE links, and IPv6 address
7018 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
7020 * .link files now understand a new Port= setting.
7022 * systemd-networkd's DHCP support gained support for DHCP option 119
7023 (domain search list).
7025 * systemd-networkd gained support for serving IPv6 address ranges using
7026 the Router Advertisement protocol. The new .network configuration
7027 section [IPv6Prefix] may be used to configure the ranges to
7028 serve. This is implemented based on a new, minimal, native server
7029 implementation of RA.
7031 * journalctl's --output= switch gained support for a new parameter
7032 "short-iso-precise" for a mode where timestamps are shown as precise
7035 * systemd-udevd's "net_id" builtin may now generate stable network
7036 interface names from IBM PowerVM VIO devices as well as ACPI platform
7039 * MulticastDNS support in systemd-resolved may now be explicitly
7040 enabled/disabled using the new MulticastDNS= configuration file
7043 * systemd-resolved may now optionally use libidn2 instead of the libidn
7044 for processing internationalized domain names. Support for libidn2
7045 should be considered experimental and should not be enabled by
7048 * "machinectl pull-tar" and related call may now do verification of
7049 downloaded images using SUSE-style .sha256 checksum files in addition
7050 to the already existing support for validating using Ubuntu-style
7053 * sd-bus gained support for a new sd_bus_message_appendv() call which
7054 is va_list equivalent of sd_bus_message_append().
7056 * sd-boot gained support for validating images using SHIM/MOK.
7058 * The SMACK code learnt support for "onlycap".
7060 * systemd-mount --umount is now much smarter in figuring out how to
7061 properly unmount a device given its mount or device path.
7063 * The code to call libnss_dns as a fallback from libnss_resolve when
7064 the communication with systemd-resolved fails was removed. This
7065 fallback was redundant and interfered with the [!UNAVAIL=return]
7066 suffix. See nss-resolve(8) for the recommended configuration.
7068 * systemd-logind may now be restarted without losing state. It stores
7069 the file descriptors for devices it manages in the system manager
7070 using the FDSTORE= mechanism. Please note that further changes in
7071 other components may be required to make use of this (for example
7072 Xorg has code to listen for stops of systemd-logind and terminate
7073 itself when logind is stopped or restarted, in order to avoid using
7074 stale file descriptors for graphical devices, which is now
7075 counterproductive and must be reverted in order for restarts of
7076 systemd-logind to be safe. See
7077 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
7079 * All kernel-install plugins are called with the environment variable
7080 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
7081 /etc/machine-id. If the machine ID could not be determined,
7082 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
7083 anything in the entry directory (passed as the second argument) if
7084 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
7085 temporary directory is passed as the entry directory and removed
7086 after all the plugins exit.
7088 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
7089 will now use its value as the machine ID instead of the machine ID
7090 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
7091 /etc/machine-info and no machine ID is set in /etc/machine-id,
7092 kernel-install will try to store the current machine ID there as
7093 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
7094 will generate a new UUID, store it in /etc/machine-info as
7095 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
7097 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
7098 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
7099 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
7100 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
7101 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
7102 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
7103 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
7104 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
7105 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
7106 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
7107 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
7108 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
7109 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
7110 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
7111 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
7112 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
7113 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
7114 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
7115 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
7116 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
7117 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
7118 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
7119 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
7120 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
7121 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
7122 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
7123 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
7124 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
7125 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
7128 — Berlin, 2017-07-12
7132 * The "hybrid" control group mode has been modified to improve
7133 compatibility with "legacy" cgroups-v1 setups. Specifically, the
7134 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
7135 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
7136 cgroups-v1 hierarchy), the only externally visible change being that
7137 the cgroups-v2 hierarchy is also mounted, to
7138 /sys/fs/cgroup/unified. This should provide a large degree of
7139 compatibility with "legacy" cgroups-v1, while taking benefit of the
7140 better management capabilities of cgroups-v2.
7142 * The default control group setup mode may be selected both a boot-time
7143 via a set of kernel command line parameters (specifically:
7144 systemd.unified_cgroup_hierarchy= and
7145 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
7146 default selected on the configure command line
7147 (--with-default-hierarchy=). The upstream default is "hybrid"
7148 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
7149 this will change in a future systemd version to be "unified" (pure
7150 cgroups-v2 mode). The third option for the compile time option is
7151 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
7152 distributions to default to "hybrid" mode for release distributions,
7153 starting with v233. We recommend "unified" for development
7154 distributions (specifically: distributions such as Fedora's rawhide)
7155 as that's where things are headed in the long run. Use "legacy" for
7156 greatest stability and compatibility only.
7158 * Note one current limitation of "unified" and "hybrid" control group
7159 setup modes: the kernel currently does not permit the systemd --user
7160 instance (i.e. unprivileged code) to migrate processes between two
7161 disconnected cgroup subtrees, even if both are managed and owned by
7162 the user. This effectively means "systemd-run --user --scope" doesn't
7163 work when invoked from outside of any "systemd --user" service or
7164 scope. Specifically, it is not supported from session scopes. We are
7165 working on fixing this in a future systemd version. (See #3388 for
7166 further details about this.)
7168 * DBus policy files are now installed into /usr rather than /etc. Make
7169 sure your system has dbus >= 1.9.18 running before upgrading to this
7170 version, or override the install path with --with-dbuspolicydir= .
7172 * All python scripts shipped with systemd (specifically: the various
7173 tests written in Python) now require Python 3.
7175 * systemd unit tests can now run standalone (without the source or
7176 build directories), and can be installed into /usr/lib/systemd/tests/
7177 with 'make install-tests'.
7179 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
7180 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
7183 * Support for the %c, %r, %R specifiers in unit files has been
7184 removed. Specifiers are not supposed to be dependent on configuration
7185 in the unit file itself (so that they resolve the same regardless
7186 where used in the unit files), but these specifiers were influenced
7187 by the Slice= option.
7189 * The shell invoked by debug-shell.service now defaults to /bin/sh in
7190 all cases. If distributions want to use a different shell for this
7191 purpose (for example Fedora's /sbin/sushell) they need to specify
7192 this explicitly at configure time using --with-debug-shell=.
7194 * The confirmation spawn prompt has been reworked to offer the
7197 (c)ontinue, proceed without asking anymore
7198 (D)ump, show the state of the unit
7199 (f)ail, don't execute the command and pretend it failed
7201 (i)nfo, show a short summary of the unit
7202 (j)obs, show jobs that are in progress
7203 (s)kip, don't execute the command and pretend it succeeded
7204 (y)es, execute the command
7206 The 'n' choice for the confirmation spawn prompt has been removed,
7207 because its meaning was confusing.
7209 The prompt may now also be redirected to an alternative console by
7210 specifying the console as parameter to systemd.confirm_spawn=.
7212 * Services of Type=notify require a READY=1 notification to be sent
7213 during startup. If no such message is sent, the service now fails,
7214 even if the main process exited with a successful exit code.
7216 * Services that fail to start up correctly now always have their
7217 ExecStopPost= commands executed. Previously, they'd enter "failed"
7218 state directly, without executing these commands.
7220 * The option MulticastDNS= of network configuration files has acquired
7221 an actual implementation. With MulticastDNS=yes a host can resolve
7222 names of remote hosts and reply to mDNS A and AAAA requests.
7224 * When units are about to be started an additional check is now done to
7225 ensure that all dependencies of type BindsTo= (when used in
7226 combination with After=) have been started.
7228 * systemd-analyze gained a new verb "syscall-filter" which shows which
7229 system call groups are defined for the SystemCallFilter= unit file
7230 setting, and which system calls they contain.
7232 * A new system call filter group "@filesystem" has been added,
7233 consisting of various file system related system calls. Group
7234 "@reboot" has been added, covering reboot, kexec and shutdown related
7235 calls. Finally, group "@swap" has been added covering swap
7236 configuration related calls.
7238 * A new unit file option RestrictNamespaces= has been added that may be
7239 used to restrict access to the various process namespace types the
7240 Linux kernel provides. Specifically, it may be used to take away the
7241 right for a service unit to create additional file system, network,
7242 user, and other namespaces. This sandboxing option is particularly
7243 relevant due to the high amount of recently discovered namespacing
7244 related vulnerabilities in the kernel.
7246 * systemd-udev's .link files gained support for a new AutoNegotiation=
7247 setting for configuring Ethernet auto-negotiation.
7249 * systemd-networkd's .network files gained support for a new
7250 ListenPort= setting in the [DHCP] section to explicitly configure the
7251 UDP client port the DHCP client shall listen on.
7253 * .network files gained a new Unmanaged= boolean setting for explicitly
7254 excluding one or more interfaces from management by systemd-networkd.
7256 * The systemd-networkd ProxyARP= option has been renamed to
7257 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
7258 renamed to ReduceARPProxy=. The old names continue to be available
7261 * systemd-networkd gained support for configuring IPv6 Proxy NDP
7262 addresses via the new IPv6ProxyNDPAddress= .network file setting.
7264 * systemd-networkd's bonding device support gained support for two new
7265 configuration options ActiveSlave= and PrimarySlave=.
7267 * The various options in the [Match] section of .network files gained
7268 support for negative matching.
7270 * New systemd-specific mount options are now understood in /etc/fstab:
7272 x-systemd.mount-timeout= may be used to configure the maximum
7273 permitted runtime of the mount command.
7275 x-systemd.device-bound may be set to bind a mount point to its
7276 backing device unit, in order to automatically remove a mount point
7277 if its backing device is unplugged. This option may also be
7278 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
7279 on the block device, which is now automatically set for all CDROM
7280 drives, so that mounted CDs are automatically unmounted when they are
7281 removed from the drive.
7283 x-systemd.after= and x-systemd.before= may be used to explicitly
7284 order a mount after or before another unit or mount point.
7286 * Enqueued start jobs for device units are now automatically garbage
7287 collected if there are no jobs waiting for them anymore.
7289 * systemctl list-jobs gained two new switches: with --after, for every
7290 queued job the jobs it's waiting for are shown; with --before the
7291 jobs which it's blocking are shown.
7293 * systemd-nspawn gained support for ephemeral boots from disk images
7294 (or in other words: --ephemeral and --image= may now be
7295 combined). Moreover, ephemeral boots are now supported for normal
7296 directories, even if the backing file system is not btrfs. Of course,
7297 if the file system does not support file system snapshots or
7298 reflinks, the initial copy operation will be relatively expensive, but
7299 this should still be suitable for many use cases.
7301 * Calendar time specifications in .timer units now support
7302 specifications relative to the end of a month by using "~" instead of
7303 "-" as separator between month and day. For example, "*-02~03" means
7304 "the third last day in February". In addition a new syntax for
7305 repeated events has been added using the "/" character. For example,
7306 "9..17/2:00" means "every two hours from 9am to 5pm".
7308 * systemd-socket-proxyd gained a new parameter --connections-max= for
7309 configuring the maximum number of concurrent connections.
7311 * sd-id128 gained a new API for generating unique IDs for the host in a
7312 way that does not leak the machine ID. Specifically,
7313 sd_id128_get_machine_app_specific() derives an ID based on the
7314 machine ID in a well-defined, non-reversible, stable way. This is
7315 useful whenever an identifier for the host is needed but where the
7316 identifier shall not be useful to identify the system beyond the
7317 scope of the application itself. (Internally this uses HMAC-SHA256 as
7318 keyed hash function using the machine ID as input.)
7320 * NotifyAccess= gained a new supported value "exec". When set
7321 notifications are accepted from all processes systemd itself invoked,
7322 including all control processes.
7324 * .nspawn files gained support for defining overlay mounts using the
7325 Overlay= and OverlayReadOnly= options. Previously this functionality
7326 was only available on the systemd-nspawn command line.
7328 * systemd-nspawn's --bind= and --overlay= options gained support for
7329 bind/overlay mounts whose source lies within the container tree by
7330 prefixing the source path with "+".
7332 * systemd-nspawn's --bind= and --overlay= options gained support for
7333 automatically allocating a temporary source directory in /var/tmp
7334 that is removed when the container dies. Specifically, if the source
7335 directory is specified as empty string this mechanism is selected. An
7336 example usage is --overlay=+/var::/var, which creates an overlay
7337 mount based on the original /var contained in the image, overlaid
7338 with a temporary directory in the host's /var/tmp. This way changes
7339 to /var are automatically flushed when the container shuts down.
7341 * systemd-nspawn --image= option does now permit raw file system block
7342 devices (in addition to images containing partition tables, as
7345 * The disk image dissection logic in systemd-nspawn gained support for
7346 automatically setting up LUKS encrypted as well as Verity protected
7347 partitions. When a container is booted from an encrypted image the
7348 passphrase is queried at start-up time. When a container with Verity
7349 data is started, the root hash is search in a ".roothash" file
7350 accompanying the disk image (alternatively, pass the root hash via
7351 the new --root-hash= command line option).
7353 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
7354 be used to dissect disk images the same way as systemd-nspawn does
7355 it, following the Bootable Partition Specification. It may even be
7356 used to mount disk images with complex partition setups (including
7357 LUKS and Verity partitions) to a local host directory, in order to
7358 inspect them. This tool is not considered public API (yet), and is
7359 thus not installed into /usr/bin. Please do not rely on its
7360 existence, since it might go away or be changed in later systemd
7363 * A new generator "systemd-verity-generator" has been added, similar in
7364 style to "systemd-cryptsetup-generator", permitting automatic setup of
7365 Verity root partitions when systemd boots up. In order to make use of
7366 this your partition setup should follow the Discoverable Partitions
7367 Specification, and the GPT partition ID of the root file system
7368 partition should be identical to the upper 128bit of the Verity root
7369 hash. The GPT partition ID of the Verity partition protecting it
7370 should be the lower 128bit of the Verity root hash. If the partition
7371 image follows this model it is sufficient to specify a single
7372 "roothash=" kernel command line argument to both configure which root
7373 image and verity partition to use as well as the root hash for
7374 it. Note that systemd-nspawn's Verity support follows the same
7375 semantics, meaning that disk images with proper Verity data in place
7376 may be booted in containers with systemd-nspawn as well as on
7377 physical systems via the verity generator. Also note that the "mkosi"
7378 tool available at https://github.com/systemd/mkosi has been updated
7379 to generate Verity protected disk images following this scheme. In
7380 fact, it has been updated to generate disk images that optionally
7381 implement a complete UEFI SecureBoot trust chain, involving a signed
7382 kernel and initrd image that incorporates such a root hash as well as
7383 a Verity-enabled root partition.
7385 * The hardware database (hwdb) udev supports has been updated to carry
7386 accelerometer quirks.
7388 * All system services are now run with a fresh kernel keyring set up
7389 for them. The invocation ID is stored by default in it, thus
7390 providing a safe, non-overridable way to determine the invocation
7393 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
7394 options for bind mounting arbitrary paths in a service-specific
7395 way. When these options are used, arbitrary host or service files and
7396 directories may be mounted to arbitrary locations in the service's
7399 * Documentation has been added that lists all of systemd's low-level
7400 environment variables:
7402 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
7404 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
7405 whether a specific socket file descriptor matches a specified socket
7408 * systemd-firstboot has been updated to check for the
7409 systemd.firstboot= kernel command line option. It accepts a boolean
7410 and when set to false the first boot questions are skipped.
7412 * systemd-fstab-generator has been updated to check for the
7413 systemd.volatile= kernel command line option, which either takes an
7414 optional boolean parameter or the special value "state". If used the
7415 system may be booted in a "volatile" boot mode. Specifically,
7416 "systemd.volatile" is used, the root directory will be mounted as
7417 tmpfs, and only /usr is mounted from the actual root file system. If
7418 "systemd.volatile=state" is used, the root directory will be mounted
7419 as usual, but /var is mounted as tmpfs. This concept provides similar
7420 functionality as systemd-nspawn's --volatile= option, but provides it
7421 on physical boots. Use this option for implementing stateless
7422 systems, or testing systems with all state and/or configuration reset
7423 to the defaults. (Note though that many distributions are not
7424 prepared to boot up without a populated /etc or /var, though.)
7426 * systemd-gpt-auto-generator gained support for LUKS encrypted root
7427 partitions. Previously it only supported LUKS encrypted partitions
7428 for all other uses, except for the root partition itself.
7430 * Socket units gained support for listening on AF_VSOCK sockets for
7431 communication in virtualized QEMU environments.
7433 * The "configure" script gained a new option --with-fallback-hostname=
7434 for specifying the fallback hostname to use if none is configured in
7435 /etc/hostname. For example, by specifying
7436 --with-fallback-hostname=fedora it is possible to default to a
7437 hostname of "fedora" on pristine installations.
7439 * systemd-cgls gained support for a new --unit= switch for listing only
7440 the control groups of a specific unit. Similar --user-unit= has been
7441 added for listing only the control groups of a specific user unit.
7443 * systemd-mount gained a new --umount switch for unmounting a mount or
7444 automount point (and all mount/automount points below it).
7446 * systemd will now refuse full configuration reloads (via systemctl
7447 daemon-reload and related calls) unless at least 16MiB of free space
7448 are available in /run. This is a safety precaution in order to ensure
7449 that generators can safely operate after the reload completed.
7451 * A new unit file option RootImage= has been added, which has a similar
7452 effect as RootDirectory= but mounts the service's root directory from
7453 a disk image instead of plain directory. This logic reuses the same
7454 image dissection and mount logic that systemd-nspawn already uses,
7455 and hence supports any disk images systemd-nspawn supports, including
7456 those following the Discoverable Partition Specification, as well as
7457 Verity enabled images. This option enables systemd to run system
7458 services directly off disk images acting as resource bundles,
7459 possibly even including full integrity data.
7461 * A new MountAPIVFS= unit file option has been added, taking a boolean
7462 argument. If enabled /proc, /sys and /dev (collectively called the
7463 "API VFS") will be mounted for the service. This is only relevant if
7464 RootDirectory= or RootImage= is used for the service, as these mounts
7465 are of course in place in the host mount namespace anyway.
7467 * systemd-nspawn gained support for a new --pivot-root= switch. If
7468 specified the root directory within the container image is pivoted to
7469 the specified mount point, while the original root disk is moved to a
7470 different place. This option enables booting of ostree images
7471 directly with systemd-nspawn.
7473 * The systemd build scripts will no longer complain if the NTP server
7474 addresses are not changed from the defaults. Google now supports
7475 these NTP servers officially. We still recommend downstreams to
7476 properly register an NTP pool with the NTP pool project though.
7478 * coredumpctl gained a new "--reverse" option for printing the list
7479 of coredumps in reverse order.
7481 * coredumpctl will now show additional information about truncated and
7482 inaccessible coredumps, as well as coredumps that are still being
7483 processed. It also gained a new --quiet switch for suppressing
7484 additional informational message in its output.
7486 * coredumpctl gained support for only showing coredumps newer and/or
7487 older than specific timestamps, using the new --since= and --until=
7488 options, reminiscent of journalctl's options by the same name.
7490 * The systemd-coredump logic has been improved so that it may be reused
7491 to collect backtraces in non-compiled languages, for example in
7492 scripting languages such as Python.
7494 * machinectl will now show the UID shift of local containers, if user
7495 namespacing is enabled for them.
7497 * systemd will now optionally run "environment generator" binaries at
7498 configuration load time. They may be used to add environment
7499 variables to the environment block passed to services invoked. One
7500 user environment generator is shipped by default that sets up
7501 environment variables based on files dropped into /etc/environment.d
7502 and ~/.config/environment.d/.
7504 * systemd-resolved now includes the new, recently published 2017 DNSSEC
7507 * hostnamed has been updated to report a new chassis type of
7508 "convertible" to cover "foldable" laptops that can both act as a
7509 tablet and as a laptop, such as various Lenovo Yoga devices.
7511 Contributions from: Adrián López, Alexander Galanin, Alexander
7512 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
7513 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
7514 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
7515 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
7516 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
7517 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
7518 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
7519 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
7520 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
7521 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
7522 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
7523 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
7524 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
7525 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
7526 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
7527 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
7528 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
7529 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
7530 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
7531 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
7532 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
7533 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
7534 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
7535 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
7536 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
7537 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
7538 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
7541 — Berlin, 2017-03-01
7545 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
7546 RestrictAddressFamilies= enabled. These sandboxing options should
7547 generally be compatible with the various external udev call-out
7548 binaries we are aware of, however there may be exceptions, in
7549 particular when exotic languages for these call-outs are used. In
7550 this case, consider turning off these settings locally.
7552 * The new RemoveIPC= option can be used to remove IPC objects owned by
7553 the user or group of a service when that service exits.
7555 * The new ProtectKernelModules= option can be used to disable explicit
7556 load and unload operations of kernel modules by a service. In
7557 addition access to /usr/lib/modules is removed if this option is set.
7559 * ProtectSystem= option gained a new value "strict", which causes the
7560 whole file system tree with the exception of /dev, /proc, and /sys,
7561 to be remounted read-only for a service.
7563 * The new ProtectKernelTunables= option can be used to disable
7564 modification of configuration files in /sys and /proc by a service.
7565 Various directories and files are remounted read-only, so access is
7566 restricted even if the file permissions would allow it.
7568 * The new ProtectControlGroups= option can be used to disable write
7569 access by a service to /sys/fs/cgroup.
7571 * Various systemd services have been hardened with
7572 ProtectKernelTunables=yes, ProtectControlGroups=yes,
7573 RestrictAddressFamilies=.
7575 * Support for dynamically creating users for the lifetime of a service
7576 has been added. If DynamicUser=yes is specified, user and group IDs
7577 will be allocated from the range 61184…65519 for the lifetime of the
7578 service. They can be resolved using the new nss-systemd.so NSS
7579 module. The module must be enabled in /etc/nsswitch.conf. Services
7580 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
7581 any resources allocated by the service will be cleaned up when the
7582 service exits. They also have ProtectHome=read-only and
7583 ProtectSystem=strict enabled, so they are not able to make any
7584 permanent modifications to the system.
7586 * The nss-systemd module also always resolves root and nobody, making
7587 it possible to have no /etc/passwd or /etc/group files in minimal
7588 container or chroot environments.
7590 * Services may be started with their own user namespace using the new
7591 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
7592 under which the service is running are mapped. All other users are
7595 * Support for the cgroup namespace has been added to systemd-nspawn. If
7596 supported by kernel, the container system started by systemd-nspawn
7597 will have its own view of the cgroup hierarchy. This new behaviour
7598 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
7600 * The new MemorySwapMax= option can be used to limit the maximum swap
7601 usage under the unified cgroup hierarchy.
7603 * Support for the CPU controller in the unified cgroup hierarchy has
7604 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
7605 options. This controller requires out-of-tree patches for the kernel
7606 and the support is provisional.
7608 * Mount and automount units may now be created transiently
7609 (i.e. dynamically at runtime via the bus API, instead of requiring
7610 unit files in the file system).
7612 * systemd-mount is a new tool which may mount file systems – much like
7613 mount(8), optionally pulling in additional dependencies through
7614 transient .mount and .automount units. For example, this tool
7615 automatically runs fsck on a backing block device before mounting,
7616 and allows the automount logic to be used dynamically from the
7617 command line for establishing mount points. This tool is particularly
7618 useful when dealing with removable media, as it will ensure fsck is
7619 run – if necessary – before the first access and that the file system
7620 is quickly unmounted after each access by utilizing the automount
7621 logic. This maximizes the chance that the file system on the
7622 removable media stays in a clean state, and if it isn't in a clean
7623 state is fixed automatically.
7625 * LazyUnmount=yes option for mount units has been added to expose the
7626 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
7629 * /efi will be used as the mount point of the EFI boot partition, if
7630 the directory is present, and the mount point was not configured
7631 through other means (e.g. fstab). If /efi directory does not exist,
7632 /boot will be used as before. This makes it easier to automatically
7633 mount the EFI partition on systems where /boot is used for something
7636 * When operating on GPT disk images for containers, systemd-nspawn will
7637 now mount the ESP to /boot or /efi according to the same rules as PID
7638 1 running on a host. This allows tools like "bootctl" to operate
7639 correctly within such containers, in order to make container images
7640 bootable on physical systems.
7642 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
7644 * Two new user session targets have been added to support running
7645 graphical sessions under the systemd --user instance:
7646 graphical-session.target and graphical-session-pre.target. See
7647 systemd.special(7) for a description of how those targets should be
7650 * The vconsole initialization code has been significantly reworked to
7651 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
7652 support unicode keymaps. Font and keymap configuration will now be
7653 copied to all allocated virtual consoles.
7655 * FreeBSD's bhyve virtualization is now detected.
7657 * Information recorded in the journal for core dumps now includes the
7658 contents of /proc/mountinfo and the command line of the process at
7659 the top of the process hierarchy (which is usually the init process
7662 * systemd-journal-gatewayd learned the --directory= option to serve
7663 files from the specified location.
7665 * journalctl --root=… can be used to peruse the journal in the
7666 /var/log/ directories inside of a container tree. This is similar to
7667 the existing --machine= option, but does not require the container to
7670 * The hardware database has been extended to support
7671 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
7674 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
7675 specify the click rate for mice which include a horizontal wheel with
7676 a click rate that is different than the one for the vertical wheel.
7678 * systemd-run gained a new --wait option that makes service execution
7679 synchronous. (Specifically, the command will not return until the
7680 specified service binary exited.)
7682 * systemctl gained a new --wait option that causes the start command to
7683 wait until the units being started have terminated again.
7685 * A new journal output mode "short-full" has been added which displays
7686 timestamps with abbreviated English day names and adds a timezone
7687 suffix. Those timestamps include more information than the default
7688 "short" output mode, and can be passed directly to journalctl's
7689 --since= and --until= options.
7691 * /etc/resolv.conf will be bind-mounted into containers started by
7692 systemd-nspawn, if possible, so any changes to resolv.conf contents
7693 are automatically propagated to the container.
7695 * The number of instances for socket-activated services originating
7696 from a single IP address can be limited with
7697 MaxConnectionsPerSource=, extending the existing setting of
7700 * systemd-networkd gained support for vcan ("Virtual CAN") interface
7703 * .netdev and .network configuration can now be extended through
7706 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
7707 Segmentation Offload, Generic Receive Offload, Large Receive Offload
7708 can be enabled and disabled using the new UDPSegmentationOffload=,
7709 TCPSegmentationOffload=, GenericSegmentationOffload=,
7710 GenericReceiveOffload=, LargeReceiveOffload= options in the
7711 [Link] section of .link files.
7713 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
7714 Port VLAN ID can be configured for bridge devices using the new STP=,
7715 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
7716 section of .netdev files.
7718 * The route table to which routes received over DHCP or RA should be
7719 added can be configured with the new RouteTable= option in the [DHCP]
7720 and [IPv6AcceptRA] sections of .network files.
7722 * The Address Resolution Protocol can be disabled on links managed by
7723 systemd-networkd using the ARP=no setting in the [Link] section of
7726 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
7727 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
7728 encode information about the result and exit codes of the current
7729 service runtime cycle.
7731 * systemd-sysctl will now configure kernel parameters in the order
7732 they occur in the configuration files. This matches what sysctl
7733 has been traditionally doing.
7735 * kernel-install "plugins" that are executed to perform various
7736 tasks after a new kernel is added and before an old one is removed
7737 can now return a special value to terminate the procedure and
7738 prevent any later plugins from running.
7740 * Journald's SplitMode=login setting has been deprecated. It has been
7741 removed from documentation, and its use is discouraged. In a future
7742 release it will be completely removed, and made equivalent to current
7743 default of SplitMode=uid.
7745 * Storage=both option setting in /etc/systemd/coredump.conf has been
7746 removed. With fast LZ4 compression storing the core dump twice is not
7749 * The --share-system systemd-nspawn option has been replaced with an
7750 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
7751 this functionality is discouraged. In addition the variables
7752 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
7753 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
7754 individual namespaces.
7756 * "machinectl list" now shows the IP address of running containers in
7757 the output, as well as OS release information.
7759 * "loginctl list" now shows the TTY of each session in the output.
7761 * sd-bus gained new API calls sd_bus_track_set_recursive(),
7762 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
7763 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
7764 tracking objects in a "recursive" mode, where a single client can be
7765 counted multiple times, if it takes multiple references.
7767 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
7768 sd_bus_get_exit_on_disconnect(). They may be used to make a
7769 process using sd-bus automatically exit if the bus connection is
7772 * Bus clients of the service manager may now "pin" loaded units into
7773 memory, by taking an explicit reference on them. This is useful to
7774 ensure the client can retrieve runtime data about the service even
7775 after the service completed execution. Taking such a reference is
7776 available only for privileged clients and should be helpful to watch
7777 running services in a race-free manner, and in particular collect
7778 information about exit statuses and results.
7780 * The nss-resolve module has been changed to strictly return UNAVAIL
7781 when communication via D-Bus with resolved failed, and NOTFOUND when
7782 a lookup completed but was negative. This means it is now possible to
7783 neatly configure fallbacks using nsswitch.conf result checking
7784 expressions. Taking benefit of this, the new recommended
7785 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
7787 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
7789 * A new setting CtrlAltDelBurstAction= has been added to
7790 /etc/systemd/system.conf which may be used to configure the precise
7791 behaviour if the user on the console presses Ctrl-Alt-Del more often
7792 than 7 times in 2s. Previously this would unconditionally result in
7793 an expedited, immediate reboot. With this new setting the precise
7794 operation may be configured in more detail, and also turned off
7797 * In .netdev files two new settings RemoteChecksumTx= and
7798 RemoteChecksumRx= are now understood that permit configuring the
7799 remote checksumming logic for VXLAN networks.
7801 * The service manager learnt a new "invocation ID" concept for invoked
7802 services. Each runtime cycle of a service will get a new invocation
7803 ID (a 128bit random UUID) assigned that identifies the current
7804 run of the service uniquely and globally. A new invocation ID
7805 is generated each time a service starts up. The journal will store
7806 the invocation ID of a service along with any logged messages, thus
7807 making the invocation ID useful for matching the online runtime of a
7808 service with the offline log data it generated in a safe way without
7809 relying on synchronized timestamps. In many ways this new service
7810 invocation ID concept is similar to the kernel's boot ID concept that
7811 uniquely and globally identifies the runtime of each boot. The
7812 invocation ID of a service is passed to the service itself via an
7813 environment variable ($INVOCATION_ID). A new bus call
7814 GetUnitByInvocationID() has been added that is similar to GetUnit()
7815 but instead of retrieving the bus path for a unit by its name
7816 retrieves it by its invocation ID. The returned path is valid only as
7817 long as the passed invocation ID is current.
7819 * systemd-resolved gained a new "DNSStubListener" setting in
7820 resolved.conf. It either takes a boolean value or the special values
7821 "udp" and "tcp", and configures whether to enable the stub DNS
7822 listener on 127.0.0.53:53.
7824 * IP addresses configured via networkd may now carry additional
7825 configuration settings supported by the kernel. New options include:
7826 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
7827 PrefixRoute=, AutoJoin=.
7829 * The PAM configuration fragment file for "user@.service" shipped with
7830 systemd (i.e. the --user instance of systemd) has been stripped to
7831 the minimum necessary to make the system boot. Previously, it
7832 contained Fedora-specific stanzas that did not apply to other
7833 distributions. It is expected that downstream distributions add
7834 additional configuration lines, matching their needs to this file,
7835 using it only as rough template of what systemd itself needs. Note
7836 that this reduced fragment does not even include an invocation of
7837 pam_limits which most distributions probably want to add, even though
7838 systemd itself does not need it. (There's also the new build time
7839 option --with-pamconfdir=no to disable installation of the PAM
7842 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
7843 capability is now also dropped from its set (in addition to
7844 CAP_SYS_MKNOD as before).
7846 * In service unit files it is now possible to connect a specific named
7847 file descriptor with stdin/stdout/stdout of an executed service. The
7848 name may be specified in matching .socket units using the
7849 FileDescriptorName= setting.
7851 * A number of journal settings may now be configured on the kernel
7852 command line. Specifically, the following options are now understood:
7853 systemd.journald.max_level_console=,
7854 systemd.journald.max_level_store=,
7855 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
7856 systemd.journald.max_level_wall=.
7858 * "systemctl is-enabled --full" will now show by which symlinks a unit
7859 file is enabled in the unit dependency tree.
7861 * Support for VeraCrypt encrypted partitions has been added to the
7862 "cryptsetup" logic and /etc/crypttab.
7864 * systemd-detect-virt gained support for a new --private-users switch
7865 that checks whether the invoking processes are running inside a user
7866 namespace. Similar, a new special value "private-users" for the
7867 existing ConditionVirtualization= setting has been added, permitting
7868 skipping of specific units in user namespace environments.
7870 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
7871 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
7872 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
7873 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
7874 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
7875 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
7876 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
7877 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
7878 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
7879 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
7880 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
7881 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
7882 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
7883 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
7884 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
7885 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
7886 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
7887 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
7888 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
7889 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
7890 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
7891 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
7892 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
7893 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
7894 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
7895 Jędrzejewski-Szmek, Zeal Jagannatha
7897 — Santa Fe, 2016-11-03
7901 * In service units the various ExecXYZ= settings have been extended
7902 with an additional special character as first argument of the
7903 assigned value: if the character '+' is used the specified command
7904 line it will be run with full privileges, regardless of User=,
7905 Group=, CapabilityBoundingSet= and similar options. The effect is
7906 similar to the existing PermissionsStartOnly= option, but allows
7907 configuration of this concept for each executed command line
7910 * Services may now alter the service watchdog timeout at runtime by
7911 sending a WATCHDOG_USEC= message via sd_notify().
7913 * MemoryLimit= and related unit settings now optionally take percentage
7914 specifications. The percentage is taken relative to the amount of
7915 physical memory in the system (or in case of containers, the assigned
7916 amount of memory). This allows scaling service resources neatly with
7917 the amount of RAM available on the system. Similarly, systemd-logind's
7918 RuntimeDirectorySize= option now also optionally takes percentage
7921 * In similar fashion TasksMax= takes percentage values now, too. The
7922 value is taken relative to the configured maximum number of processes
7923 on the system. The per-service task maximum has been changed to 15%
7924 using this functionality. (Effectively this is an increase of 512 →
7925 4915 for service units, given the kernel's default pid_max setting.)
7927 * Calendar time specifications in .timer units now understand a ".."
7928 syntax for time ranges. Example: "4..7:10" may now be used for
7929 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
7932 * The InaccessableDirectories=, ReadOnlyDirectories= and
7933 ReadWriteDirectories= unit file settings have been renamed to
7934 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
7935 applied to all kinds of file nodes, and not just directories, with
7936 the exception of symlinks. Specifically these settings may now be
7937 used on block and character device nodes, UNIX sockets and FIFOS as
7938 well as regular files. The old names of these settings remain
7939 available for compatibility.
7941 * systemd will now log about all service processes it kills forcibly
7942 (using SIGKILL) because they remained after the clean shutdown phase
7943 of the service completed. This should help identifying services that
7944 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
7945 systemd-logind's configuration a similar log message is generated for
7946 processes killed at the end of each session due to this setting.
7948 * systemd will now set the $JOURNAL_STREAM environment variable for all
7949 services whose stdout/stderr are connected to the Journal (which
7950 effectively means by default: all services). The variable contains
7951 the device and inode number of the file descriptor used for
7952 stdout/stderr. This may be used by invoked programs to detect whether
7953 their stdout/stderr is connected to the Journal, in which case they
7954 can switch over to direct Journal communication, thus being able to
7955 pass extended, structured metadata along with their log messages. As
7956 one example, this is now used by glib's logging primitives.
7958 * When using systemd's default tmp.mount unit for /tmp, the mount point
7959 will now be established with the "nosuid" and "nodev" options. This
7960 avoids privilege escalation attacks that put traps and exploits into
7961 /tmp. However, this might cause problems if you e. g. put container
7962 images or overlays into /tmp; if you need this, override tmp.mount's
7963 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
7966 * systemd now supports the "memory" cgroup controller also on
7969 * The systemd-cgtop tool now optionally takes a control group path as
7970 command line argument. If specified, the control group list shown is
7971 limited to subgroups of that group.
7973 * The SystemCallFilter= unit file setting gained support for
7974 pre-defined, named system call filter sets. For example
7975 SystemCallFilter=@clock is now an effective way to make all clock
7976 changing-related system calls unavailable to a service. A number of
7977 similar pre-defined groups are defined. Writing system call filters
7978 for system services is simplified substantially with this new
7979 concept. Accordingly, all of systemd's own, long-running services now
7980 enable system call filtering based on this, by default.
7982 * A new service setting MemoryDenyWriteExecute= has been added, taking
7983 a boolean value. If turned on, a service may no longer create memory
7984 mappings that are writable and executable at the same time. This
7985 enhances security for services where this is enabled as it becomes
7986 harder to dynamically write and then execute memory in exploited
7987 service processes. This option has been enabled for all of systemd's
7988 own long-running services.
7990 * A new RestrictRealtime= service setting has been added, taking a
7991 boolean argument. If set the service's processes may no longer
7992 acquire realtime scheduling. This improves security as realtime
7993 scheduling may otherwise be used to easily freeze the system.
7995 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
7996 value. This may be used for requesting that the system manager inside
7997 of the container reports start-up completion to nspawn which then
7998 propagates this notification further to the service manager
7999 supervising nspawn itself. A related option NotifyReady= in .nspawn
8000 files has been added too. This functionality allows ordering of the
8001 start-up of multiple containers using the usual systemd ordering
8004 * machinectl gained a new command "stop" that is an alias for
8007 * systemd-resolved gained support for contacting DNS servers on
8008 link-local IPv6 addresses.
8010 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
8011 its caches. A method call for requesting the same operation has been
8012 added to the bus API too, and is made available via "systemd-resolve
8015 * systemd-resolve gained a new --status switch. If passed a brief
8016 summary of the used DNS configuration with per-interface information
8019 * resolved.conf gained a new Cache= boolean option, defaulting to
8020 on. If turned off local DNS caching is disabled. This comes with a
8021 performance penalty in particular when DNSSEC is enabled. Note that
8022 resolved disables its internal caching implicitly anyway, when the
8023 configured DNS server is on a host-local IP address such as ::1 or
8024 127.0.0.1, thus automatically avoiding double local caching.
8026 * systemd-resolved now listens on the local IP address 127.0.0.53:53
8027 for DNS requests. This improves compatibility with local programs
8028 that do not use the libc NSS or systemd-resolved's bus APIs for name
8029 resolution. This minimal DNS service is only available to local
8030 programs and does not implement the full DNS protocol, but enough to
8031 cover local DNS clients. A new, static resolv.conf file, listing just
8032 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
8033 now recommended to make /etc/resolv.conf a symlink to this file in
8034 order to route all DNS lookups to systemd-resolved, regardless if
8035 done via NSS, the bus API or raw DNS packets. Note that this local
8036 DNS service is not as fully featured as the libc NSS or
8037 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
8038 used to deliver link-local address information (as this implies
8039 sending a local interface index along), LLMNR/mDNS support via this
8040 interface is severely restricted. It is thus strongly recommended for
8041 all applications to use the libc NSS API or native systemd-resolved
8044 * systemd-networkd's bridge support learned a new setting
8045 VLANFiltering= for controlling VLAN filtering. Moreover a new section
8046 in .network files has been added for configuring VLAN bridging in
8047 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
8049 * systemd-networkd's IPv6 Router Advertisement code now makes use of
8050 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
8051 now be acquired without relying on DHCPv6. Two new options
8052 UseDomains= and UseDNS= have been added to configure this behaviour.
8054 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
8055 renamed IPv6AcceptRA=, without altering its behaviour. The old
8056 setting name remains available for compatibility reasons.
8058 * The systemd-networkd VTI/VTI6 tunneling support gained new options
8059 Key=, InputKey= and OutputKey=.
8061 * systemd-networkd gained support for VRF ("Virtual Routing Function")
8062 interface configuration.
8064 * "systemctl edit" may now be used to create new unit files by
8065 specifying the --force switch.
8067 * sd-event gained a new function sd_event_get_iteration() for
8068 requesting the current iteration counter of the event loop. It starts
8069 at zero and is increased by one with each event loop iteration.
8071 * A new rpm macro %systemd_ordering is provided by the macros.systemd
8072 file. It can be used in lieu of %systemd_requires in packages which
8073 don't use any systemd functionality and are intended to be installed
8074 in minimal containers without systemd present. This macro provides
8075 ordering dependencies to ensure that if the package is installed in
8076 the same rpm transaction as systemd, systemd will be installed before
8077 the scriptlets for the package are executed, allowing unit presets
8080 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
8081 been added to simplify packaging of generators.
8083 * The os-release file gained VERSION_CODENAME field for the
8084 distribution nickname (e.g. VERSION_CODENAME=woody).
8086 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
8087 can be set to disable parsing of metadata and the creation
8088 of persistent symlinks for that device.
8090 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
8091 to make them available to logged-in users has been reverted.
8093 * Much of the common code of the various systemd components is now
8094 built into an internal shared library libsystemd-shared-231.so
8095 (incorporating the systemd version number in the name, to be updated
8096 with future releases) that the components link to. This should
8097 decrease systemd footprint both in memory during runtime and on
8098 disk. Note that the shared library is not for public use, and is
8099 neither API nor ABI stable, but is likely to change with every new
8100 released update. Packagers need to make sure that binaries
8101 linking to libsystemd-shared.so are updated in step with the
8104 * Configuration for "mkosi" is now part of the systemd
8105 repository. mkosi is a tool to easily build legacy-free OS images,
8106 and is available on github: https://github.com/systemd/mkosi. If
8107 "mkosi" is invoked in the build tree a new raw OS image is generated
8108 incorporating the systemd sources currently being worked on and a
8109 clean, fresh distribution installation. The generated OS image may be
8110 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
8111 UEFI PC. This functionality is particularly useful to easily test
8112 local changes made to systemd in a pristine, defined environment. See
8113 doc/HACKING for details.
8115 * configure learned the --with-support-url= option to specify the
8116 distribution's bugtracker.
8118 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
8119 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
8120 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
8121 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
8122 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
8123 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
8124 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
8125 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
8126 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
8127 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
8128 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
8129 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
8130 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
8131 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
8132 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
8133 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
8134 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
8135 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
8136 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
8138 — Berlin, 2016-07-25
8142 * DNSSEC is now turned on by default in systemd-resolved (in
8143 "allow-downgrade" mode), but may be turned off during compile time by
8144 passing "--with-default-dnssec=no" to "configure" (and of course,
8145 during runtime with DNSSEC= in resolved.conf). We recommend
8146 downstreams to leave this on at least during development cycles and
8147 report any issues with the DNSSEC logic upstream. We are very
8148 interested in collecting feedback about the DNSSEC validator and its
8149 limitations in the wild. Note however, that DNSSEC support is
8150 probably nothing downstreams should turn on in stable distros just
8151 yet, as it might create incompatibilities with a few DNS servers and
8152 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
8153 automatically whenever we detect such incompatible setups, but there
8154 might be systems we do not cover yet. Hence: please help us testing
8155 the DNSSEC code, leave this on where you can, report back, but then
8156 again don't consider turning this on in your stable, LTS or
8157 production release just yet. (Note that you have to enable
8158 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
8159 and its DNSSEC mode for hostname resolution from local
8162 * systemd-resolve conveniently resolves DANE records with the --tlsa
8163 option and OPENPGPKEY records with the --openpgp option. It also
8164 supports dumping raw DNS record data via the new --raw= switch.
8166 * systemd-logind will now by default terminate user processes that are
8167 part of the user session scope unit (session-XX.scope) when the user
8168 logs out. This behavior is controlled by the KillUserProcesses=
8169 setting in logind.conf, and the previous default of "no" is now
8170 changed to "yes". This means that user sessions will be properly
8171 cleaned up after, but additional steps are necessary to allow
8172 intentionally long-running processes to survive logout.
8174 While the user is logged in at least once, user@.service is running,
8175 and any service that should survive the end of any individual login
8176 session can be started at a user service or scope using systemd-run.
8177 systemd-run(1) man page has been extended with an example which shows
8178 how to run screen in a scope unit underneath user@.service. The same
8179 command works for tmux.
8181 After the user logs out of all sessions, user@.service will be
8182 terminated too, by default, unless the user has "lingering" enabled.
8183 To effectively allow users to run long-term tasks even if they are
8184 logged out, lingering must be enabled for them. See loginctl(1) for
8185 details. The default polkit policy was modified to allow users to
8186 set lingering for themselves without authentication.
8188 Previous defaults can be restored at compile time by the
8189 --without-kill-user-processes option to "configure".
8191 * systemd-logind gained new configuration settings SessionsMax= and
8192 InhibitorsMax=, both with a default of 8192. It will not register new
8193 user sessions or inhibitors above this limit.
8195 * systemd-logind will now reload configuration on SIGHUP.
8197 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
8198 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
8199 enable. Also, support for the "io" cgroup controller in the unified
8200 hierarchy has been added, so that the "memory", "pids" and "io" are
8201 now the controllers that are supported on the unified hierarchy.
8203 WARNING: it is not possible to use previous systemd versions with
8204 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
8205 is necessary to also update systemd in the initramfs if using the
8206 unified hierarchy. An updated SELinux policy is also required.
8208 * LLDP support has been extended, and both passive (receive-only) and
8209 active (sender) modes are supported. Passive mode ("routers-only") is
8210 enabled by default in systemd-networkd. Active LLDP mode is enabled
8211 by default for containers on the internal network. The "networkctl
8212 lldp" command may be used to list information gathered. "networkctl
8213 status" will also show basic LLDP information on connected peers now.
8215 * The IAID and DUID unique identifier sent in DHCP requests may now be
8216 configured for the system and each .network file managed by
8217 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
8219 * systemd-networkd gained support for configuring proxy ARP support for
8220 each interface, via the ProxyArp= setting in .network files. It also
8221 gained support for configuring the multicast querier feature of
8222 bridge devices, via the new MulticastQuerier= setting in .netdev
8223 files. Similarly, snooping on the IGMP traffic can be controlled
8224 via the new setting MulticastSnooping=.
8226 A new setting PreferredLifetime= has been added for addresses
8227 configured in .network file to configure the lifetime intended for an
8230 The systemd-networkd DHCP server gained the option EmitRouter=, which
8231 defaults to yes, to configure whether the DHCP Option 3 (Router)
8234 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
8235 systemd-socket-activate and installed into /usr/bin. It is now fully
8238 * systemd-journald now uses separate threads to flush changes to disk
8239 when closing journal files, thus reducing impact of slow disk I/O on
8240 logging performance.
8242 * The sd-journal API gained two new calls
8243 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
8244 can be used to open journal files using file descriptors instead of
8245 file or directory paths. sd_journal_open_container() has been
8246 deprecated, sd_journal_open_directory_fd() should be used instead
8247 with the flag SD_JOURNAL_OS_ROOT.
8249 * journalctl learned a new output mode "-o short-unix" that outputs log
8250 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
8251 UTC). It also gained support for a new --no-hostname setting to
8252 suppress the hostname column in the family of "short" output modes.
8254 * systemd-ask-password now optionally skips printing of the password to
8255 stdout with --no-output which can be useful in scripts.
8257 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
8258 (devices tagged with ID_MAKER_TOOL) are now tagged with
8259 "uaccess" and are available to logged in users.
8261 * The DeviceAllow= unit setting now supports specifiers (with "%").
8263 * "systemctl show" gained a new --value switch, which allows print a
8264 only the contents of a specific unit property, without also printing
8265 the property's name. Similar support was added to "show*" verbs
8266 of loginctl and machinectl that output "key=value" lists.
8268 * A new unit type "generated" was added for files dynamically generated
8269 by generator tools. Similarly, a new unit type "transient" is used
8270 for unit files created using the runtime API. "systemctl enable" will
8271 refuse to operate on such files.
8273 * A new command "systemctl revert" has been added that may be used to
8274 revert to the vendor version of a unit file, in case local changes
8275 have been made by adding drop-ins or overriding the unit file.
8277 * "machinectl clean" gained a new verb to automatically remove all or
8278 just hidden container images.
8280 * systemd-tmpfiles gained support for a new line type "e" for emptying
8281 directories, if they exist, without creating them if they don't.
8283 * systemd-nspawn gained support for automatically patching the UID/GIDs
8284 of the owners and the ACLs of all files and directories in a
8285 container tree to match the UID/GID user namespacing range selected
8286 for the container invocation. This mode is enabled via the new
8287 --private-users-chown switch. It also gained support for
8288 automatically choosing a free, previously unused UID/GID range when
8289 starting a container, via the new --private-users=pick setting (which
8290 implies --private-users-chown). Together, these options for the first
8291 time make user namespacing for nspawn containers fully automatic and
8292 thus deployable. The systemd-nspawn@.service template unit file has
8293 been changed to use this functionality by default.
8295 * systemd-nspawn gained a new --network-zone= switch, that allows
8296 creating ad-hoc virtual Ethernet links between multiple containers,
8297 that only exist as long as at least one container referencing them is
8298 running. This allows easy connecting of multiple containers with a
8299 common link that implements an Ethernet broadcast domain. Each of
8300 these network "zones" may be named relatively freely by the user, and
8301 may be referenced by any number of containers, but each container may
8302 only reference one of these "zones". On the lower level, this is
8303 implemented by an automatically managed bridge network interface for
8304 each zone, that is created when the first container referencing its
8305 zone is created and removed when the last one referencing its zone
8308 * The default start timeout may now be configured on the kernel command
8309 line via systemd.default_timeout_start_sec=. It was already
8310 configurable via the DefaultTimeoutStartSec= option in
8311 /etc/systemd/system.conf.
8313 * Socket units gained a new TriggerLimitIntervalSec= and
8314 TriggerLimitBurst= setting to configure a limit on the activation
8315 rate of the socket unit.
8317 * The LimitNICE= setting now optionally takes normal UNIX nice values
8318 in addition to the raw integer limit value. If the specified
8319 parameter is prefixed with "+" or "-" and is in the range -20…19 the
8320 value is understood as UNIX nice value. If not prefixed like this it
8321 is understood as raw RLIMIT_NICE limit.
8323 * Note that the effect of the PrivateDevices= unit file setting changed
8324 slightly with this release: the per-device /dev file system will be
8325 mounted read-only from this version on, and will have "noexec"
8326 set. This (minor) change of behavior might cause some (exceptional)
8327 legacy software to break, when PrivateDevices=yes is set for its
8328 service. Please leave PrivateDevices= off if you run into problems
8331 * systemd-bootchart has been split out to a separate repository:
8332 https://github.com/systemd/systemd-bootchart
8334 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
8335 merged into the kernel in its current form.
8337 * The compatibility libraries libsystemd-daemon.so,
8338 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
8339 which have been deprecated since systemd-209 have been removed along
8340 with the corresponding pkg-config files. All symbols provided by
8341 those libraries are provided by libsystemd.so.
8343 * The Capabilities= unit file setting has been removed (it is ignored
8344 for backwards compatibility). AmbientCapabilities= and
8345 CapabilityBoundingSet= should be used instead.
8347 * A new special target has been added, initrd-root-device.target,
8348 which creates a synchronization point for dependencies of the root
8349 device in early userspace. Initramfs builders must ensure that this
8350 target is now included in early userspace.
8352 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
8353 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
8354 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
8355 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
8356 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
8357 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
8358 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
8359 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
8360 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
8361 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
8362 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
8363 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
8364 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
8365 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
8366 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
8367 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
8368 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
8369 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
8370 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
8371 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
8372 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
8373 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
8374 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
8375 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
8376 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
8379 — Fairfax, 2016-05-21
8383 * The systemd-resolved DNS resolver service has gained a substantial
8384 set of new features, most prominently it may now act as a DNSSEC
8385 validating stub resolver. DNSSEC mode is currently turned off by
8386 default, but is expected to be turned on by default in one of the
8387 next releases. For now, we invite everybody to test the DNSSEC logic
8388 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
8389 service also gained a full set of D-Bus interfaces, including calls
8390 to configure DNS and DNSSEC settings per link (for use by external
8391 network management software). systemd-resolved and systemd-networkd
8392 now distinguish between "search" and "routing" domains. The former
8393 are used to qualify single-label names, the latter are used purely
8394 for routing lookups within certain domains to specific links.
8395 resolved now also synthesizes RRs for all entries from /etc/hosts.
8397 * The systemd-resolve tool (which is a client utility for
8398 systemd-resolved) has been improved considerably and is now fully
8399 supported and documented. Hence it has moved from /usr/lib/systemd to
8402 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
8405 * The coredump collection logic has been reworked: when a coredump is
8406 collected it is now written to disk, compressed and processed
8407 (including stacktrace extraction) from a new instantiated service
8408 systemd-coredump@.service, instead of directly from the
8409 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
8410 processing large coredumps can take up a substantial amount of
8411 resources and time, and this previously happened entirely outside of
8412 systemd's service supervision. With the new logic the core_pattern
8413 hook only does minimal metadata collection before passing off control
8414 to the new instantiated service, which is configured with a time
8415 limit, a nice level and other settings to minimize negative impact on
8416 the rest of the system. Also note that the new logic will honour the
8417 RLIMIT_CORE setting of the crashed process, which now allows users
8418 and processes to turn off coredumping for their processes by setting
8421 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
8422 and all forked processes by default. Previously, PID 1 would leave
8423 the setting at "0" for all processes, as set by the kernel. Note that
8424 the resource limit traditionally has no effect on the generated
8425 coredumps on the system if the /proc/sys/kernel/core_pattern hook
8426 logic is used. Since the limit is now honoured (see above) its
8427 default has been changed so that the coredumping logic is enabled by
8428 default for all processes, while allowing specific opt-out.
8430 * When the stacktrace is extracted from processes of system users, this
8431 is now done as "systemd-coredump" user, in order to sandbox this
8432 potentially security sensitive parsing operation. (Note that when
8433 processing coredumps of normal users this is done under the user ID
8434 of process that crashed, as before.) Packagers should take notice
8435 that it is now necessary to create the "systemd-coredump" system user
8436 and group at package installation time.
8438 * The systemd-activate socket activation testing tool gained support
8439 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
8440 and --seqpacket switches. It also has been extended to support both
8441 new-style and inetd-style file descriptor passing. Use the new
8442 --inetd switch to request inetd-style file descriptor passing.
8444 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
8445 variable, which takes a boolean value. If set to false, ANSI color
8446 output is disabled in the tools even when run on a terminal that
8449 * The VXLAN support in networkd now supports two new settings
8450 DestinationPort= and PortRange=.
8452 * A new systemd.machine_id= kernel command line switch has been added,
8453 that may be used to set the machine ID in /etc/machine-id if it is
8454 not initialized yet. This command line option has no effect if the
8455 file is already initialized.
8457 * systemd-nspawn gained a new --as-pid2 switch that invokes any
8458 specified command line as PID 2 rather than PID 1 in the
8459 container. In this mode PID 1 is a minimal stub init process that
8460 implements the special POSIX and Linux semantics of PID 1 regarding
8461 signal and child process management. Note that this stub init process
8462 is implemented in nspawn itself and requires no support from the
8463 container image. This new logic is useful to support running
8464 arbitrary commands in the container, as normal processes are
8465 generally not prepared to run as PID 1.
8467 * systemd-nspawn gained a new --chdir= switch for setting the current
8468 working directory for the process started in the container.
8470 * "journalctl /dev/sda" will now output all kernel log messages for
8471 specified device from the current boot, in addition to all devices
8472 that are parents of it. This should make log output about devices
8473 pretty useful, as long as kernel drivers attach enough metadata to
8474 the log messages. (The usual SATA drivers do.)
8476 * The sd-journal API gained two new calls
8477 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
8478 that report whether log data from /run or /var has been found.
8480 * journalctl gained a new switch "--fields" that prints all journal
8481 record field names currently in use in the journal. This is backed
8482 by two new sd-journal API calls sd_journal_enumerate_fields() and
8483 sd_journal_restart_fields().
8485 * Most configurable timeouts in systemd now expect an argument of
8486 "infinity" to turn them off, instead of "0" as before. The semantics
8487 from now on is that a timeout of "0" means "now", and "infinity"
8488 means "never". To maintain backwards compatibility, "0" continues to
8489 turn off previously existing timeout settings.
8491 * "systemctl reload-or-try-restart" has been renamed to "systemctl
8492 try-reload-or-restart" to clarify what it actually does: the "try"
8493 logic applies to both reloading and restarting, not just restarting.
8494 The old name continues to be accepted for compatibility.
8496 * On boot-up, when PID 1 detects that the system clock is behind the
8497 release date of the systemd version in use, the clock is now set
8498 to the latter. Previously, this was already done in timesyncd, in order
8499 to avoid running with clocks set to the various clock epochs such as
8500 1902, 1938 or 1970. With this change the logic is now done in PID 1
8501 in addition to timesyncd during early boot-up, so that it is enforced
8502 before the first process is spawned by systemd. Note that the logic
8503 in timesyncd remains, as it is more comprehensive and ensures
8504 clock monotonicity by maintaining a persistent timestamp file in
8505 /var. Since /var is generally not available in earliest boot or the
8506 initrd, this part of the logic remains in timesyncd, and is not done
8509 * Support for tweaking details in net_cls.class_id through the
8510 NetClass= configuration directive has been removed, as the kernel
8511 people have decided to deprecate that controller in cgroup v2.
8512 Userspace tools such as nftables are moving over to setting rules
8513 that are specific to the full cgroup path of a task, which obsoletes
8514 these controllers anyway. The NetClass= directive is kept around for
8515 legacy compatibility reasons. For a more in-depth description of the
8516 kernel change, please refer to the respective upstream commit:
8518 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
8520 * A new service setting RuntimeMaxSec= has been added that may be used
8521 to specify a maximum runtime for a service. If the timeout is hit, the
8522 service is terminated and put into a failure state.
8524 * A new service setting AmbientCapabilities= has been added. It allows
8525 configuration of additional Linux process capabilities that are
8526 passed to the activated processes. This is only available on very
8529 * The process resource limit settings in service units may now be used
8530 to configure hard and soft limits individually.
8532 * The various libsystemd APIs such as sd-bus or sd-event now publicly
8533 expose support for gcc's __attribute__((cleanup())) C extension.
8534 Specifically, for many object destructor functions alternative
8535 versions have been added that have names suffixed with "p" and take a
8536 pointer to a pointer to the object to destroy, instead of just a
8537 pointer to the object itself. This is useful because these destructor
8538 functions may be used directly as parameters to the cleanup
8539 construct. Internally, systemd has been a heavy user of this GCC
8540 extension for a long time, and with this change similar support is
8541 now available to consumers of the library outside of systemd. Note
8542 that by using this extension in your sources compatibility with old
8543 and strictly ANSI compatible C compilers is lost. However, all gcc or
8544 LLVM versions of recent years support this extension.
8546 * Timer units gained support for a new setting RandomizedDelaySec= that
8547 allows configuring some additional randomized delay to the configured
8548 time. This is useful to spread out timer events to avoid load peaks in
8549 clusters or larger setups.
8551 * Calendar time specifications now support sub-second accuracy.
8553 * Socket units now support listening on SCTP and UDP-lite protocol
8556 * The sd-event API now comes with a full set of man pages.
8558 * Older versions of systemd contained experimental support for
8559 compressing journal files and coredumps with the LZ4 compressor that
8560 was not compatible with the lz4 binary (due to API limitations of the
8561 lz4 library). This support has been removed; only support for files
8562 compatible with the lz4 binary remains. This LZ4 logic is now
8563 officially supported and no longer considered experimental.
8565 * The dkr image import logic has been removed again from importd. dkr's
8566 micro-services focus doesn't fit into the machine image focus of
8567 importd, and quickly got out of date with the upstream dkr API.
8569 * Creation of the /run/lock/lockdev/ directory was dropped from
8570 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
8571 been available for many years. If you still need this, you need to
8572 create your own tmpfiles.d config file with:
8574 d /run/lock/lockdev 0775 root lock -
8576 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
8577 and RebootArgument= have been moved from the [Service] section of
8578 unit files to [Unit], and they are now supported on all unit types,
8579 not just service units. Of course, systemd will continue to
8580 understand these settings also at the old location, in order to
8581 maintain compatibility.
8583 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
8584 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
8585 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
8586 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
8587 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
8588 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
8589 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
8590 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
8591 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
8592 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
8593 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
8594 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
8595 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
8596 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
8597 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
8598 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
8599 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
8600 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
8601 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8603 — Berlin, 2016-02-11
8607 * A number of properties previously only settable in unit
8608 files are now also available as properties to set when
8609 creating transient units programmatically via the bus, as it
8610 is exposed with systemd-run's --property=
8611 setting. Specifically, these are: SyslogIdentifier=,
8612 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
8613 EnvironmentFile=, ReadWriteDirectories=,
8614 ReadOnlyDirectories=, InaccessibleDirectories=,
8615 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
8617 * When creating transient services via the bus API it is now
8618 possible to pass in a set of file descriptors to use as
8619 STDIN/STDOUT/STDERR for the invoked process.
8621 * Slice units may now be created transiently via the bus APIs,
8622 similar to the way service and scope units may already be
8623 created transiently.
8625 * Wherever systemd expects a calendar timestamp specification
8626 (like in journalctl's --since= and --until= switches) UTC
8627 timestamps are now supported. Timestamps suffixed with "UTC"
8628 are now considered to be in Universal Time Coordinated
8629 instead of the local timezone. Also, timestamps may now
8630 optionally be specified with sub-second accuracy. Both of
8631 these additions also apply to recurring calendar event
8632 specification, such as OnCalendar= in timer units.
8634 * journalctl gained a new "--sync" switch that asks the
8635 journal daemon to write all so far unwritten log messages to
8636 disk and sync the files, before returning.
8638 * systemd-tmpfiles learned two new line types "q" and "Q" that
8639 operate like "v", but also set up a basic btrfs quota
8640 hierarchy when used on a btrfs file system with quota
8643 * tmpfiles' "v", "q" and "Q" will now create a plain directory
8644 instead of a subvolume (even on a btrfs file system) if the
8645 root directory is a plain directory, and not a
8646 subvolume. This should simplify things with certain chroot()
8647 environments which are not aware of the concept of btrfs
8650 * systemd-detect-virt gained a new --chroot switch to detect
8651 whether execution takes place in a chroot() environment.
8653 * CPUAffinity= now takes CPU index ranges in addition to
8656 * The various memory-related resource limit settings (such as
8657 LimitAS=) now understand the usual K, M, G, … suffixes to
8658 the base of 1024 (IEC). Similar, the time-related resource
8659 limit settings understand the usual min, h, day, … suffixes
8662 * There's a new system.conf setting DefaultTasksMax= to
8663 control the default TasksMax= setting for services and
8664 scopes running on the system. (TasksMax= is the primary
8665 setting that exposes the "pids" cgroup controller on systemd
8666 and was introduced in the previous systemd release.) The
8667 setting now defaults to 512, which means services that are
8668 not explicitly configured otherwise will only be able to
8669 create 512 processes or threads at maximum, from this
8670 version on. Note that this means that thread- or
8671 process-heavy services might need to be reconfigured to set
8672 TasksMax= to a higher value. It is sufficient to set
8673 TasksMax= in these specific unit files to a higher value, or
8674 even "infinity". Similar, there's now a logind.conf setting
8675 UserTasksMax= that defaults to 4096 and limits the total
8676 number of processes or tasks each user may own
8677 concurrently. nspawn containers also have the TasksMax=
8678 value set by default now, to 8192. Note that all of this
8679 only has an effect if the "pids" cgroup controller is
8680 enabled in the kernel. The general benefit of these changes
8681 should be a more robust and safer system, that provides a
8682 certain amount of per-service fork() bomb protection.
8684 * systemd-nspawn gained the new --network-veth-extra= switch
8685 to define additional and arbitrarily-named virtual Ethernet
8686 links between the host and the container.
8688 * A new service execution setting PassEnvironment= has been
8689 added that allows importing select environment variables
8690 from PID1's environment block into the environment block of
8693 * Timer units gained support for a new RemainAfterElapse=
8694 setting which takes a boolean argument. It defaults to on,
8695 exposing behaviour unchanged to previous releases. If set to
8696 off, timer units are unloaded after they elapsed if they
8697 cannot elapse again. This is particularly useful for
8698 transient timer units, which shall not stay around longer
8699 than until they first elapse.
8701 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
8702 default now (the kernel default is 16). This is beneficial
8703 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
8704 allows substantially larger numbers of queued
8705 datagrams. This should increase the capability of systemd to
8706 parallelize boot-up, as logging and sd_notify() are unlikely
8707 to stall execution anymore. If you need to change the value
8708 from the new defaults, use the usual sysctl.d/ snippets.
8710 * The compression framing format used by the journal or
8711 coredump processing has changed to be in line with what the
8712 official LZ4 tools generate. LZ4 compression support in
8713 systemd was considered unsupported previously, as the format
8714 was not compatible with the normal tools. With this release
8715 this has changed now, and it is hence safe for downstream
8716 distributions to turn it on. While not compressing as well
8717 as the XZ, LZ4 is substantially faster, which makes
8718 it a good default choice for the compression logic in the
8719 journal and in coredump handling.
8721 * Any reference to /etc/mtab has been dropped from
8722 systemd. The file has been obsolete since a while, but
8723 systemd refused to work on systems where it was incorrectly
8724 set up (it should be a symlink or non-existent). Please make
8725 sure to update to util-linux 2.27.1 or newer in conjunction
8726 with this systemd release, which also drops any reference to
8727 /etc/mtab. If you maintain a distribution make sure that no
8728 software you package still references it, as this is a
8729 likely source of bugs. There's also a glibc bug pending,
8730 asking for removal of any reference to this obsolete file:
8732 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
8734 Note that only util-linux versions built with
8735 --enable-libmount-force-mountinfo are supported.
8737 * Support for the ".snapshot" unit type has been removed. This
8738 feature turned out to be little useful and little used, and
8739 has now been removed from the core and from systemctl.
8741 * The dependency types RequiresOverridable= and
8742 RequisiteOverridable= have been removed from systemd. They
8743 have been used only very sparingly to our knowledge and
8744 other options that provide a similar effect (such as
8745 systemctl --mode=ignore-dependencies) are much more useful
8746 and commonly used. Moreover, they were only half-way
8747 implemented as the option to control behaviour regarding
8748 these dependencies was never added to systemctl. By removing
8749 these dependency types the execution engine becomes a bit
8750 simpler. Unit files that use these dependencies should be
8751 changed to use the non-Overridable dependency types
8752 instead. In fact, when parsing unit files with these
8753 options, that's what systemd will automatically convert them
8754 too, but it will also warn, asking users to fix the unit
8755 files accordingly. Removal of these dependency types should
8756 only affect a negligible number of unit files in the wild.
8758 * Behaviour of networkd's IPForward= option changed
8759 (again). It will no longer maintain a per-interface setting,
8760 but propagate one way from interfaces where this is enabled
8761 to the global kernel setting. The global setting will be
8762 enabled when requested by a network that is set up, but
8763 never be disabled again. This change was made to make sure
8764 IPv4 and IPv6 behaviour regarding packet forwarding is
8765 similar (as the Linux IPv6 stack does not support
8766 per-interface control of this setting) and to minimize
8769 * In unit files the behaviour of %u, %U, %h, %s has
8770 changed. These specifiers will now unconditionally resolve
8771 to the various user database fields of the user that the
8772 systemd instance is running as, instead of the user
8773 configured in the specific unit via User=. Note that this
8774 effectively doesn't change much, as resolving of these
8775 specifiers was already turned off in the --system instance
8776 of systemd, as we cannot do NSS lookups from PID 1. In the
8777 --user instance of systemd these specifiers where correctly
8778 resolved, but hardly made any sense, since the user instance
8779 lacks privileges to do user switches anyway, and User= is
8780 hence useless. Moreover, even in the --user instance of
8781 systemd behaviour was awkward as it would only take settings
8782 from User= assignment placed before the specifier into
8783 account. In order to unify and simplify the logic around
8784 this the specifiers will now always resolve to the
8785 credentials of the user invoking the manager (which in case
8786 of PID 1 is the root user).
8788 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
8789 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
8790 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
8791 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
8792 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
8793 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
8794 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
8795 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
8796 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
8797 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
8798 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
8799 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
8800 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
8801 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
8804 — Berlin, 2015-11-18
8808 * systemd now depends on util-linux v2.27. More specifically,
8809 the newly added mount monitor feature in libmount now
8810 replaces systemd's former own implementation.
8812 * libmount mandates /etc/mtab not to be regular file, and
8813 systemd now enforces this condition at early boot.
8814 /etc/mtab has been deprecated and warned about for a very
8815 long time, so systems running systemd should already have
8816 stopped having this file around as anything else than a
8817 symlink to /proc/self/mounts.
8819 * Support for the "pids" cgroup controller has been added. It
8820 allows accounting the number of tasks in a cgroup and
8821 enforcing limits on it. This adds two new setting
8822 TasksAccounting= and TasksMax= to each unit, as well as a
8823 global option DefaultTasksAccounting=.
8825 * Support for the "net_cls" cgroup controller has been added.
8826 It allows assigning a net class ID to each task in the
8827 cgroup, which can then be used in firewall rules and traffic
8828 shaping configurations. Note that the kernel netfilter net
8829 class code does not currently work reliably for ingress
8830 packets on unestablished sockets.
8832 This adds a new config directive called NetClass= to CGroup
8833 enabled units. Allowed values are positive numbers for fixed
8834 assignments and "auto" for picking a free value
8837 * 'systemctl is-system-running' now returns 'offline' if the
8838 system is not booted with systemd. This command can now be
8839 used as a substitute for 'systemd-notify --booted'.
8841 * Watchdog timeouts have been increased to 3 minutes for all
8842 in-tree service files. Apparently, disk IO issues are more
8843 frequent than we hoped, and user reported >1 minute waiting
8846 * 'machine-id-commit' functionality has been merged into
8847 'machine-id-setup --commit'. The separate binary has been
8850 * The WorkingDirectory= directive in unit files may now be set
8851 to the special value '~'. In this case, the working
8852 directory is set to the home directory of the user
8853 configured in User=.
8855 * "machinectl shell" will now open the shell in the home
8856 directory of the selected user by default.
8858 * The CrashChVT= configuration file setting is renamed to
8859 CrashChangeVT=, following our usual logic of not
8860 abbreviating unnecessarily. The old directive is still
8861 supported for compat reasons. Also, this directive now takes
8862 an integer value between 1 and 63, or a boolean value. The
8863 formerly supported '-1' value for disabling stays around for
8866 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
8867 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
8868 RootDirectory= properties can now be set for transient
8871 * The systemd-analyze tool gained a new "set-log-target" verb
8872 to change the logging target the system manager logs to
8873 dynamically during runtime. This is similar to how
8874 "systemd-analyze set-log-level" already changes the log
8877 * In nspawn /sys is now mounted as tmpfs, with only a selected
8878 set of subdirectories mounted in from the real sysfs. This
8879 enhances security slightly, and is useful for ensuring user
8880 namespaces work correctly.
8882 * Support for USB FunctionFS activation has been added. This
8883 allows implementation of USB gadget services that are
8884 activated as soon as they are requested, so that they don't
8885 have to run continuously, similar to classic socket
8888 * The "systemctl exit" command now optionally takes an
8889 additional parameter that sets the exit code to return from
8890 the systemd manager when exiting. This is only relevant when
8891 running the systemd user instance, or when running the
8892 system instance in a container.
8894 * sd-bus gained the new API calls sd_bus_path_encode_many()
8895 and sd_bus_path_decode_many() that allow easy encoding and
8896 decoding of multiple identifier strings inside a D-Bus
8897 object path. Another new call sd_bus_default_flush_close()
8898 has been added to flush and close per-thread default
8901 * systemd-cgtop gained support for a -M/--machine= switch to
8902 show the control groups within a certain container only.
8904 * "systemctl kill" gained support for an optional --fail
8905 switch. If specified the requested operation will fail of no
8906 processes have been killed, because the unit had no
8907 processes attached, or similar.
8909 * A new systemd.crash_reboot=1 kernel command line option has
8910 been added that triggers a reboot after crashing. This can
8911 also be set through CrashReboot= in systemd.conf.
8913 * The RuntimeDirectory= setting now understands unit
8914 specifiers like %i or %f.
8916 * A new (still internal) library API sd-ipv4acd has been added,
8917 that implements address conflict detection for IPv4. It's
8918 based on code from sd-ipv4ll, and will be useful for
8919 detecting DHCP address conflicts.
8921 * File descriptors passed during socket activation may now be
8922 named. A new API sd_listen_fds_with_names() is added to
8923 access the names. The default names may be overridden,
8924 either in the .socket file using the FileDescriptorName=
8925 parameter, or by passing FDNAME= when storing the file
8926 descriptors using sd_notify().
8928 * systemd-networkd gained support for:
8930 - Setting the IPv6 Router Advertisement settings via
8931 IPv6AcceptRouterAdvertisements= in .network files.
8933 - Configuring the HelloTimeSec=, MaxAgeSec= and
8934 ForwardDelaySec= bridge parameters in .netdev files.
8936 - Configuring PreferredSource= for static routes in
8939 * The "ask-password" framework used to query for LUKS harddisk
8940 passwords or SSL passwords during boot gained support for
8941 caching passwords in the kernel keyring, if it is
8942 available. This makes sure that the user only has to type in
8943 a passphrase once if there are multiple objects to unlock
8944 with the same one. Previously, such password caching was
8945 available only when Plymouth was used; this moves the
8946 caching logic into the systemd codebase itself. The
8947 "systemd-ask-password" utility gained a new --keyname=
8948 switch to control which kernel keyring key to use for
8949 caching a password in. This functionality is also useful for
8950 enabling display managers such as gdm to automatically
8951 unlock the user's GNOME keyring if its passphrase, the
8952 user's password and the harddisk password are the same, if
8953 gdm-autologin is used.
8955 * When downloading tar or raw images using "machinectl
8956 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
8957 file is now also downloaded, if it is available and stored
8958 next to the image file.
8960 * Units of type ".socket" gained a new boolean setting
8961 Writable= which is only useful in conjunction with
8962 ListenSpecial=. If true, enables opening the specified
8963 special file in O_RDWR mode rather than O_RDONLY mode.
8965 * systemd-rfkill has been reworked to become a singleton
8966 service that is activated through /dev/rfkill on each rfkill
8967 state change and saves the settings to disk. This way,
8968 systemd-rfkill is now compatible with devices that exist
8969 only intermittendly, and even restores state if the previous
8970 system shutdown was abrupt rather than clean.
8972 * The journal daemon gained support for vacuuming old journal
8973 files controlled by the number of files that shall remain,
8974 in addition to the already existing control by size and by
8975 date. This is useful as journal interleaving performance
8976 degrades with too many separate journal files, and allows
8977 putting an effective limit on them. The new setting defaults
8978 to 100, but this may be changed by setting SystemMaxFiles=
8979 and RuntimeMaxFiles= in journald.conf. Also, the
8980 "journalctl" tool gained the new --vacuum-files= switch to
8981 manually vacuum journal files to leave only the specified
8982 number of files in place.
8984 * udev will now create /dev/disk/by-path links for ATA devices
8985 on kernels where that is supported.
8987 * Galician, Serbian, Turkish and Korean translations were added.
8989 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
8990 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
8991 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
8992 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
8993 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
8994 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
8995 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
8996 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
8997 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
8998 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
8999 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
9000 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
9001 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
9002 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
9003 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
9004 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
9005 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
9006 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
9008 — Berlin, 2015-10-07
9012 * The DHCP implementation of systemd-networkd gained a set of
9015 - The DHCP server now supports emitting DNS and NTP
9016 information. It may be enabled and configured via
9017 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
9018 and NTP information is enabled, but no servers are
9019 configured, the corresponding uplink information (if there
9020 is any) is propagated.
9022 - Server and client now support transmission and reception
9023 of timezone information. It can be configured via the
9024 newly introduced network options UseTimezone=,
9025 EmitTimezone=, and Timezone=. Transmission of timezone
9026 information is enabled between host and containers by
9027 default now: the container will change its local timezone
9028 to what the host has set.
9030 - Lease timeouts can now be configured via
9031 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
9033 - The DHCP server improved on the stability of
9034 leases. Clients are more likely to get the same lease
9035 information back, even if the server loses state.
9037 - The DHCP server supports two new configuration options to
9038 control the lease address pool metrics, PoolOffset= and
9041 * The encapsulation limit of tunnels in systemd-networkd may
9042 now be configured via 'EncapsulationLimit='. It allows
9043 modifying the maximum additional levels of encapsulation
9044 that are permitted to be prepended to a packet.
9046 * systemd now supports the concept of user buses replacing
9047 session buses, if used with dbus-1.10 (and enabled via dbus
9048 --enable-user-session). It previously only supported this on
9049 kdbus-enabled systems, and this release expands this to
9050 'dbus-daemon' systems.
9052 * systemd-networkd now supports predictable interface names
9055 * systemd now optionally supports the new Linux kernel
9056 "unified" control group hierarchy. If enabled via the kernel
9057 command-line option 'systemd.unified_cgroup_hierarchy=1',
9058 systemd will try to mount the unified cgroup hierarchy
9059 directly on /sys/fs/cgroup. If not enabled, or not
9060 available, systemd will fall back to the legacy cgroup
9061 hierarchy setup, as before. Host system and containers can
9062 mix and match legacy and unified hierarchies as they
9063 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
9064 environment variable to individually select the hierarchy to
9065 use for executed containers. By default, nspawn will use the
9066 unified hierarchy for the containers if the host uses the
9067 unified hierarchy, and the legacy hierarchy otherwise.
9068 Please note that at this point the unified hierarchy is an
9069 experimental kernel feature and is likely to change in one
9070 of the next kernel releases. Therefore, it should not be
9071 enabled by default in downstream distributions yet. The
9072 minimum required kernel version for the unified hierarchy to
9073 work is 4.2. Note that when the unified hierarchy is used
9074 for the first time delegated access to controllers is
9075 safe. Because of this systemd-nspawn containers will get
9076 access to controllers now, as will systemd user
9077 sessions. This means containers and user sessions may now
9078 manage their own resources, partitioning up what the system
9081 * A new special scope unit "init.scope" has been introduced
9082 that encapsulates PID 1 of the system. It may be used to
9083 determine resource usage and enforce resource limits on PID
9084 1 itself. PID 1 hence moved out of the root of the control
9087 * The cgtop tool gained support for filtering out kernel
9088 threads when counting tasks in a control group. Also, the
9089 count of processes is now recursively summed up by
9090 default. Two options -k and --recursive= have been added to
9091 revert to old behaviour. The tool has also been updated to
9092 work correctly in containers now.
9094 * systemd-nspawn's --bind= and --bind-ro= options have been
9095 extended to allow creation of non-recursive bind mounts.
9097 * libsystemd gained two new calls sd_pid_get_cgroup() and
9098 sd_peer_get_cgroup() which return the control group path of
9099 a process or peer of a connected AF_UNIX socket. This
9100 function call is particularly useful when implementing
9101 delegated subtrees support in the control group hierarchy.
9103 * The "sd-event" event loop API of libsystemd now supports
9104 correct dequeuing of real-time signals, without losing
9107 * When systemd requests a polkit decision when managing units it
9108 will now add additional fields to the request, including unit
9109 name and desired operation. This enables more powerful polkit
9110 policies, that make decisions depending on these parameters.
9112 * nspawn learnt support for .nspawn settings files, that may
9113 accompany the image files or directories of containers, and
9114 may contain additional settings for the container. This is
9115 an alternative to configuring container parameters via the
9116 nspawn command line.
9118 Contributions from: Cristian Rodríguez, Daniel Mack, David
9119 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
9120 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
9121 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
9122 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
9123 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
9124 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
9125 Andersen, Tom Gundersen, Torstein Husebø
9127 — Berlin, 2015-09-08
9131 * machinectl gained a new verb 'shell' which opens a fresh
9132 shell on the target container or the host. It is similar to
9133 the existing 'login' command of machinectl, but spawns the
9134 shell directly without prompting for username or
9135 password. The pseudo machine '.host' now refers to the local
9136 host and is used by default. Hence, 'machinectl shell' can
9137 be used as replacement for 'su -' which spawns a session as
9138 a fresh systemd unit in a way that is fully isolated from
9139 the originating session.
9141 * systemd-networkd learned to cope with private-zone DHCP
9142 options and allows other programs to query the values.
9144 * SELinux access control when enabling/disabling units is no
9145 longer enforced with this release. The previous implementation
9146 was incorrect, and a new corrected implementation is not yet
9147 available. As unit file operations are still protected via
9148 polkit and D-Bus policy this is not a security problem. Yet,
9149 distributions which care about optimal SELinux support should
9150 probably not stabilize on this release.
9152 * sd-bus gained support for matches of type "arg0has=", that
9153 test for membership of strings in string arrays sent in bus
9156 * systemd-resolved now dumps the contents of its DNS and LLMNR
9157 caches to the logs on reception of the SIGUSR1 signal. This
9158 is useful to debug DNS behaviour.
9160 * The coredumpctl tool gained a new --directory= option to
9161 operate on journal files in a specific directory.
9163 * "systemctl reboot" and related commands gained a new
9164 "--message=" option which may be used to set a free-text
9165 wall message when shutting down or rebooting the
9166 system. This message is also logged, which is useful for
9167 figuring out the reason for a reboot or shutdown a
9170 * The "systemd-resolve-host" tool's -i switch now takes
9171 network interface numbers as alternative to interface names.
9173 * A new unit file setting for services has been introduced:
9174 UtmpMode= allows configuration of how precisely systemd
9175 handles utmp and wtmp entries for the service if this is
9176 enabled. This allows writing services that appear similar to
9177 user sessions in the output of the "w", "who", "last" and
9180 * systemd-resolved will now locally synthesize DNS resource
9181 records for the "localhost" and "gateway" domains as well as
9182 the local hostname. This should ensure that clients querying
9183 RRs via resolved will get similar results as those going via
9184 NSS, if nss-myhostname is enabled.
9186 Contributions from: Alastair Hughes, Alex Crawford, Daniel
9187 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
9188 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
9189 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
9190 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
9191 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
9192 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
9193 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
9194 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
9195 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
9196 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
9197 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
9199 — Berlin, 2015-08-27
9203 * The systemd-efi-boot-generator functionality was merged into
9204 systemd-gpt-auto-generator.
9206 * systemd-networkd now supports Group Policy for vxlan
9207 devices. It can be enabled via the new boolean configuration
9208 option called 'GroupPolicyExtension='.
9210 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
9211 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
9212 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
9214 — Berlin, 2015-07-31
9218 * The python-systemd code has been removed from the systemd repository.
9219 A new repository has been created which accommodates the code from
9220 now on, and we kindly ask distributions to create a separate package
9221 for this: https://github.com/systemd/python-systemd
9223 * The systemd daemon will now reload its main configuration
9224 (/etc/systemd/system.conf) on daemon-reload.
9226 * sd-dhcp now exposes vendor specific extensions via
9227 sd_dhcp_lease_get_vendor_specific().
9229 * systemd-networkd gained a number of new configuration options.
9231 - A new boolean configuration option for TAP devices called
9232 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
9233 device, thus allowing to send and receive GSO packets.
9235 - A new tunnel configuration option called 'CopyDSCP='.
9236 If enabled, the DSCP field of ip6 tunnels is copied into the
9237 decapsulated packet.
9239 - A set of boolean bridge configuration options were added.
9240 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
9241 and 'UnicastFlood=' are now parsed by networkd and applied to the
9242 respective bridge link device via the respective IFLA_BRPORT_*
9245 - A new string configuration option to override the hostname sent
9246 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
9247 is true, networkd will use the configured hostname instead of the
9248 system hostname when sending DHCP requests.
9250 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
9251 networkd will configure the IPv6 flow-label of the tunnel device
9252 according to RFC2460.
9254 - The 'macvtap' virtual network devices are now supported, similar to
9255 the already supported 'macvlan' devices.
9257 * systemd-resolved now implements RFC5452 to improve resilience against
9258 cache poisoning. Additionally, source port randomization is enabled
9259 by default to further protect against DNS spoofing attacks.
9261 * nss-mymachines now supports translating UIDs and GIDs of running
9262 containers with user-namespaces enabled. If a container 'foo'
9263 translates a host uid 'UID' to the container uid 'TUID', then
9264 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
9265 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
9266 mapped as 'vg-foo-TGID'.
9268 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
9269 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
9270 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
9271 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
9272 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
9273 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
9274 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
9275 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
9276 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
9277 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
9279 — Berlin, 2015-07-29
9283 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
9284 There are no known issues with current sysfs, and udev does not need
9285 or should be used to work around such bugs.
9287 * udev does no longer enable USB HID power management. Several reports
9288 indicate, that some devices cannot handle that setting.
9290 * The udev accelerometer helper was removed. The functionality
9291 is now fully included in iio-sensor-proxy. But this means,
9292 older iio-sensor-proxy versions will no longer provide
9293 accelerometer/orientation data with this systemd version.
9294 Please upgrade iio-sensor-proxy to version 1.0.
9296 * networkd gained a new configuration option IPv6PrivacyExtensions=
9297 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
9298 for Stateless Address") on selected networks.
9300 * For the sake of fewer build-time dependencies and less code in the
9301 main repository, the python bindings are about to be removed in the
9302 next release. A new repository has been created which accommodates
9303 the code from now on, and we kindly ask distributions to create a
9304 separate package for this. The removal will take place in v223.
9306 https://github.com/systemd/python-systemd
9308 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
9309 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
9310 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
9311 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
9312 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
9313 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
9314 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
9315 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
9316 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
9317 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
9319 — Berlin, 2015-07-07
9323 * The sd-bus.h and sd-event.h APIs have now been declared
9324 stable and have been added to the official interface of
9325 libsystemd.so. sd-bus implements an alternative D-Bus client
9326 library, that is relatively easy to use, very efficient and
9327 supports both classic D-Bus as well as kdbus as transport
9328 backend. sd-event is a generic event loop abstraction that
9329 is built around Linux epoll, but adds features such as event
9330 prioritization or efficient timer handling. Both APIs are good
9331 choices for C programs looking for a bus and/or event loop
9332 implementation that is minimal and does not have to be
9333 portable to other kernels.
9335 * kdbus support is no longer compile-time optional. It is now
9336 always built-in. However, it can still be disabled at
9337 runtime using the kdbus=0 kernel command line setting, and
9338 that setting may be changed to default to off, by specifying
9339 --disable-kdbus at build-time. Note though that the kernel
9340 command line setting has no effect if the kdbus.ko kernel
9341 module is not installed, in which case kdbus is (obviously)
9342 also disabled. We encourage all downstream distributions to
9343 begin testing kdbus by adding it to the kernel images in the
9344 development distributions, and leaving kdbus support in
9347 * The minimal required util-linux version has been bumped to
9350 * Support for chkconfig (--enable-chkconfig) was removed in
9351 favor of calling an abstraction tool
9352 /lib/systemd/systemd-sysv-install. This needs to be
9353 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
9354 in README for details.
9356 * If there's a systemd unit and a SysV init script for the
9357 same service name, and the user executes "systemctl enable"
9358 for it (or a related call), then this will now enable both
9359 (or execute the related operation on both), not just the
9362 * The libudev API documentation has been converted from gtkdoc
9365 * gudev has been removed from the systemd tree, it is now an
9368 * The systemd-cgtop tool learnt a new --raw switch to generate
9369 "raw" (machine parsable) output.
9371 * networkd's IPForwarding= .network file setting learnt the
9372 new setting "kernel", which ensures that networkd does not
9373 change the IP forwarding sysctl from the default kernel
9376 * The systemd-logind bus API now exposes a new boolean
9377 property "Docked" that reports whether logind considers the
9378 system "docked", i.e. connected to a docking station or not.
9380 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
9381 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
9382 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
9383 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
9384 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
9385 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
9386 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
9387 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
9388 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
9389 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
9390 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
9391 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
9392 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
9393 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
9394 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
9395 Fink, Zbigniew Jędrzejewski-Szmek
9397 — Berlin, 2015-06-19
9401 * The gudev library has been extracted into a separate repository
9402 available at: https://git.gnome.org/browse/libgudev/
9403 It is now managed as part of the Gnome project. Distributions
9404 are recommended to pass --disable-gudev to systemd and use
9405 gudev from the Gnome project instead. gudev is still included
9406 in systemd, for now. It will be removed soon, though. Please
9407 also see the announcement-thread on systemd-devel:
9408 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
9410 * systemd now exposes a CPUUsageNSec= property for each
9411 service unit on the bus, that contains the overall consumed
9412 CPU time of a service (the sum of what each process of the
9413 service consumed). This value is only available if
9414 CPUAccounting= is turned on for a service, and is then shown
9415 in the "systemctl status" output.
9417 * Support for configuring alternative mappings of the old SysV
9418 runlevels to systemd targets has been removed. They are now
9419 hardcoded in a way that runlevels 2, 3, 4 all map to
9420 multi-user.target and 5 to graphical.target (which
9421 previously was already the default behaviour).
9423 * The auto-mounter logic gained support for mount point
9424 expiry, using a new TimeoutIdleSec= setting in .automount
9425 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
9427 * The EFI System Partition (ESP) as mounted to /boot by
9428 systemd-efi-boot-generator will now be unmounted
9429 automatically after 2 minutes of not being used. This should
9430 minimize the risk of ESP corruptions.
9432 * New /etc/fstab options x-systemd.requires= and
9433 x-systemd.requires-mounts-for= are now supported to express
9434 additional dependencies for mounts. This is useful for
9435 journaling file systems that support external journal
9436 devices or overlay file systems that require underlying file
9437 systems to be mounted.
9439 * systemd does not support direct live-upgrades (via systemctl
9440 daemon-reexec) from versions older than v44 anymore. As no
9441 distribution we are aware of shipped such old versions in a
9442 stable release this should not be problematic.
9444 * When systemd forks off a new per-connection service instance
9445 it will now set the $REMOTE_ADDR environment variable to the
9446 remote IP address, and $REMOTE_PORT environment variable to
9447 the remote IP port. This behaviour is similar to the
9448 corresponding environment variables defined by CGI.
9450 * systemd-networkd gained support for uplink failure
9451 detection. The BindCarrier= option allows binding interface
9452 configuration dynamically to the link sense of other
9453 interfaces. This is useful to achieve behaviour like in
9456 * systemd-networkd gained support for configuring the DHCP
9457 client identifier to use when requesting leases.
9459 * systemd-networkd now has a per-network UseNTP= option to
9460 configure whether NTP server information acquired via DHCP
9461 is passed on to services like systemd-timesyncd.
9463 * systemd-networkd gained support for vti6 tunnels.
9465 * Note that systemd-networkd manages the sysctl variable
9466 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
9467 it is configured for since v219. The variable controls IP
9468 forwarding, and is a per-interface alternative to the global
9469 /proc/sys/net/ipv[46]/ip_forward. This setting is
9470 configurable in the IPForward= option, which defaults to
9471 "no". This means if networkd is used for an interface it is
9472 no longer sufficient to set the global sysctl option to turn
9473 on IP forwarding! Instead, the .network file option
9474 IPForward= needs to be turned on! Note that the
9475 implementation of this behaviour was broken in v219 and has
9478 * Many bonding and vxlan options are now configurable in
9481 * systemd-nspawn gained a new --property= setting to set unit
9482 properties for the container scope. This is useful for
9483 setting resource parameters (e.g. "CPUShares=500") on
9484 containers started from the command line.
9486 * systemd-nspawn gained a new --private-users= switch to make
9487 use of user namespacing available on recent Linux kernels.
9489 * systemd-nspawn may now be called as part of a shell pipeline
9490 in which case the pipes used for stdin and stdout are passed
9491 directly to the process invoked in the container, without
9492 indirection via a pseudo tty.
9494 * systemd-nspawn gained a new switch to control the UNIX
9495 signal to use when killing the init process of the container
9498 * systemd-nspawn gained a new --overlay= switch for mounting
9499 overlay file systems into the container using the new kernel
9502 * When a container image is imported via systemd-importd and
9503 the host file system is not btrfs, a loopback block device
9504 file is created in /var/lib/machines.raw with a btrfs file
9505 system inside. It is then mounted to /var/lib/machines to
9506 enable btrfs features for container management. The loopback
9507 file and btrfs file system is grown as needed when container
9508 images are imported via systemd-importd.
9510 * systemd-machined/systemd-importd gained support for btrfs
9511 quota, to enforce container disk space limits on disk. This
9512 is exposed in "machinectl set-limit".
9514 * systemd-importd now can import containers from local .tar,
9515 .raw and .qcow2 images, and export them to .tar and .raw. It
9516 can also import dkr v2 images now from the network (on top
9519 * systemd-importd gained support for verifying downloaded
9520 images with gpg2 (previously only gpg1 was supported).
9522 * systemd-machined, systemd-logind, systemd: most bus calls are
9523 now accessible to unprivileged processes via polkit. Also,
9524 systemd-logind will now allow users to kill their own sessions
9525 without further privileges or authorization.
9527 * systemd-shutdownd has been removed. This service was
9528 previously responsible for implementing scheduled shutdowns
9529 as exposed in /usr/bin/shutdown's time parameter. This
9530 functionality has now been moved into systemd-logind and is
9531 accessible via a bus interface.
9533 * "systemctl reboot" gained a new switch --firmware-setup that
9534 can be used to reboot into the EFI firmware setup, if that
9535 is available. systemd-logind now exposes an API on the bus
9536 to trigger such reboots, in case graphical desktop UIs want
9537 to cover this functionality.
9539 * "systemctl enable", "systemctl disable" and "systemctl mask"
9540 now support a new "--now" switch. If specified the units
9541 that are enabled will also be started, and the ones
9542 disabled/masked also stopped.
9544 * The Gummiboot EFI boot loader tool has been merged into
9545 systemd, and renamed to "systemd-boot". The bootctl tool has been
9546 updated to support systemd-boot.
9548 * An EFI kernel stub has been added that may be used to create
9549 kernel EFI binaries that contain not only the actual kernel,
9550 but also an initrd, boot splash, command line and OS release
9551 information. This combined binary can then be signed as a
9552 single image, so that the firmware can verify it all in one
9553 step. systemd-boot has special support for EFI binaries created
9554 like this and can extract OS release information from them
9555 and show them in the boot menu. This functionality is useful
9556 to implement cryptographically verified boot schemes.
9558 * Optional support has been added to systemd-fsck to pass
9559 fsck's progress report to an AF_UNIX socket in the file
9562 * udev will no longer create device symlinks for all block devices by
9563 default. A deny list for excluding special block devices from this
9564 logic has been turned into an allow list that requires picking block
9565 devices explicitly that require device symlinks.
9567 * A new (currently still internal) API sd-device.h has been
9568 added to libsystemd. This modernized API is supposed to
9569 replace libudev eventually. In fact, already much of libudev
9570 is now just a wrapper around sd-device.h.
9572 * A new hwdb database for storing metadata about pointing
9573 stick devices has been added.
9575 * systemd-tmpfiles gained support for setting file attributes
9576 similar to the "chattr" tool with new 'h' and 'H' lines.
9578 * systemd-journald will no longer unconditionally set the
9579 btrfs NOCOW flag on new journal files. This is instead done
9580 with tmpfiles snippet using the new 'h' line type. This
9581 allows easy disabling of this logic, by masking the
9582 journal-nocow.conf tmpfiles file.
9584 * systemd-journald will now translate audit message types to
9585 human readable identifiers when writing them to the
9586 journal. This should improve readability of audit messages.
9588 * The LUKS logic gained support for the offset= and skip=
9589 options in /etc/crypttab, as previously implemented by
9592 * /usr/lib/os-release gained a new optional field VARIANT= for
9593 distributions that support multiple variants (such as a
9594 desktop edition, a server edition, …)
9596 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
9597 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
9598 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
9599 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
9600 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
9601 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
9602 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
9603 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
9604 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
9605 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
9606 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
9607 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
9608 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
9609 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
9610 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
9611 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
9612 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
9613 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
9614 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
9615 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
9616 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
9617 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
9618 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
9619 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
9620 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
9621 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
9622 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
9624 — Berlin, 2015-05-22
9628 * Introduce a new API "sd-hwdb.h" for querying the hardware
9629 metadata database. With this minimal interface one can query
9630 and enumerate the udev hwdb, decoupled from the old libudev
9631 library. libudev's interface for this is now only a wrapper
9632 around sd-hwdb. A new tool systemd-hwdb has been added to
9633 interface with and update the database.
9635 * When any of systemd's tools copies files (for example due to
9636 tmpfiles' C lines) a btrfs reflink will attempted first,
9637 before bytewise copying is done.
9639 * systemd-nspawn gained a new --ephemeral switch. When
9640 specified a btrfs snapshot is taken of the container's root
9641 directory, and immediately removed when the container
9642 terminates again. Thus, a container can be started whose
9643 changes never alter the container's root directory, and are
9644 lost on container termination. This switch can also be used
9645 for starting a container off the root file system of the
9646 host without affecting the host OS. This switch is only
9647 available on btrfs file systems.
9649 * systemd-nspawn gained a new --template= switch. It takes the
9650 path to a container tree to use as template for the tree
9651 specified via --directory=, should that directory be
9652 missing. This allows instantiating containers dynamically,
9653 on first run. This switch is only available on btrfs file
9656 * When a .mount unit refers to a mount point on which multiple
9657 mounts are stacked, and the .mount unit is stopped all of
9658 the stacked mount points will now be unmounted until no
9659 mount point remains.
9661 * systemd now has an explicit notion of supported and
9662 unsupported unit types. Jobs enqueued for unsupported unit
9663 types will now fail with an "unsupported" error code. More
9664 specifically .swap, .automount and .device units are not
9665 supported in containers, .busname units are not supported on
9666 non-kdbus systems. .swap and .automount are also not
9667 supported if their respective kernel compile time options
9670 * machinectl gained support for two new "copy-from" and
9671 "copy-to" commands for copying files from a running
9672 container to the host or vice versa.
9674 * machinectl gained support for a new "bind" command to bind
9675 mount host directories into local containers. This is
9676 currently only supported for nspawn containers.
9678 * networkd gained support for configuring bridge forwarding
9679 database entries (fdb) from .network files.
9681 * A new tiny daemon "systemd-importd" has been added that can
9682 download container images in tar, raw, qcow2 or dkr formats,
9683 and make them available locally in /var/lib/machines, so
9684 that they can run as nspawn containers. The daemon can GPG
9685 verify the downloads (not supported for dkr, since it has no
9686 provisions for verifying downloads). It will transparently
9687 decompress bz2, xz, gzip compressed downloads if necessary,
9688 and restore sparse files on disk. The daemon uses privilege
9689 separation to ensure the actual download logic runs with
9690 fewer privileges than the daemon itself. machinectl has
9691 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
9692 make the functionality of importd available to the
9693 user. With this in place the Fedora and Ubuntu "Cloud"
9694 images can be downloaded and booted as containers unmodified
9695 (the Fedora images lack the appropriate GPG signature files
9696 currently, so they cannot be verified, but this will change
9697 soon, hopefully). Note that downloading images is currently
9698 only fully supported on btrfs.
9700 * machinectl is now able to list container images found in
9701 /var/lib/machines, along with some metadata about sizes of
9702 disk and similar. If the directory is located on btrfs and
9703 quota is enabled, this includes quota display. A new command
9704 "image-status" has been added that shows additional
9705 information about images.
9707 * machinectl is now able to clone container images
9708 efficiently, if the underlying file system (btrfs) supports
9709 it, with the new "machinectl clone" command. It also
9710 gained commands for renaming and removing images, as well as
9711 marking them read-only or read-write (supported also on
9712 legacy file systems).
9714 * networkd gained support for collecting LLDP network
9715 announcements, from hardware that supports this. This is
9716 shown in networkctl output.
9718 * systemd-run gained support for a new -t (--pty) switch for
9719 invoking a binary on a pty whose input and output is
9720 connected to the invoking terminal. This allows executing
9721 processes as system services while interactively
9722 communicating with them via the terminal. Most interestingly
9723 this is supported across container boundaries. Invoking
9724 "systemd-run -t /bin/bash" is an alternative to running a
9725 full login session, the difference being that the former
9726 will not register a session, nor go through the PAM session
9729 * tmpfiles gained support for a new "v" line type for creating
9730 btrfs subvolumes. If the underlying file system is a legacy
9731 file system, this automatically degrades to creating a
9732 normal directory. Among others /var/lib/machines is now
9733 created like this at boot, should it be missing.
9735 * The directory /var/lib/containers/ has been deprecated and
9736 been replaced by /var/lib/machines. The term "machines" has
9737 been used in the systemd context as generic term for both
9738 VMs and containers, and hence appears more appropriate for
9739 this, as the directory can also contain raw images bootable
9742 * systemd-nspawn when invoked with -M but without --directory=
9743 or --image= is now capable of searching for the container
9744 root directory, subvolume or disk image automatically, in
9745 /var/lib/machines. systemd-nspawn@.service has been updated
9746 to make use of this, thus allowing it to be used for raw
9749 * A new machines.target unit has been introduced that is
9750 supposed to group all containers/VMs invoked as services on
9751 the system. systemd-nspawn@.service has been updated to
9752 integrate with that.
9754 * machinectl gained a new "start" command, for invoking a
9755 container as a service. "machinectl start foo" is mostly
9756 equivalent to "systemctl start systemd-nspawn@foo.service",
9757 but handles escaping in a nicer way.
9759 * systemd-nspawn will now mount most of the cgroupfs tree
9760 read-only into each container, with the exception of the
9761 container's own subtree in the name=systemd hierarchy.
9763 * journald now sets the special FS_NOCOW file flag for its
9764 journal files. This should improve performance on btrfs, by
9765 avoiding heavy fragmentation when journald's write-pattern
9766 is used on COW file systems. It degrades btrfs' data
9767 integrity guarantees for the files to the same levels as for
9768 ext3/ext4 however. This should be OK though as journald does
9769 its own data integrity checks and all its objects are
9770 checksummed on disk. Also, journald should handle btrfs disk
9771 full events a lot more gracefully now, by processing SIGBUS
9772 errors, and not relying on fallocate() anymore.
9774 * When journald detects that journal files it is writing to
9775 have been deleted it will immediately start new journal
9778 * systemd now provides a way to store file descriptors
9779 per-service in PID 1. This is useful for daemons to ensure
9780 that fds they require are not lost during a daemon
9781 restart. The fds are passed to the daemon on the next
9782 invocation in the same way socket activation fds are
9783 passed. This is now used by journald to ensure that the
9784 various sockets connected to all the system's stdout/stderr
9785 are not lost when journald is restarted. File descriptors
9786 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
9787 an extension to sd_notify(). Note that a limit is enforced
9788 on the number of fds a service can store in PID 1, and it
9789 defaults to 0, so that no fds may be stored, unless this is
9790 explicitly turned on.
9792 * The default TERM variable to use for units connected to a
9793 terminal, when no other value is explicitly is set is now
9794 vt220 rather than vt102. This should be fairly safe still,
9795 but allows PgUp/PgDn work.
9797 * The /etc/crypttab option header= as known from Debian is now
9800 * "loginctl user-status" and "loginctl session-status" will
9801 now show the last 10 lines of log messages of the
9802 user/session following the status output. Similar,
9803 "machinectl status" will show the last 10 log lines
9804 associated with a virtual machine or container
9805 service. (Note that this is usually not the log messages
9806 done in the VM/container itself, but simply what the
9807 container manager logs. For nspawn this includes all console
9810 * "loginctl session-status" without further argument will now
9811 show the status of the session of the caller. Similar,
9812 "lock-session", "unlock-session", "activate",
9813 "enable-linger", "disable-linger" may now be called without
9814 session/user parameter in which case they apply to the
9815 caller's session/user.
9817 * An X11 session scriptlet is now shipped that uploads
9818 $DISPLAY and $XAUTHORITY into the environment of the systemd
9819 --user daemon if a session begins. This should improve
9820 compatibility with X11 enabled applications run as systemd
9823 * Generators are now subject to masking via /etc and /run, the
9824 same way as unit files.
9826 * networkd .network files gained support for configuring
9827 per-link IPv4/IPv6 packet forwarding as well as IPv4
9828 masquerading. This is by default turned on for veth links to
9829 containers, as registered by systemd-nspawn. This means that
9830 nspawn containers run with --network-veth will now get
9831 automatic routed access to the host's networks without any
9832 further configuration or setup, as long as networkd runs on
9835 * systemd-nspawn gained the --port= (-p) switch to expose TCP
9836 or UDP posts of a container on the host. With this in place
9837 it is possible to run containers with private veth links
9838 (--network-veth), and have their functionality exposed on
9839 the host as if their services were running directly on the
9842 * systemd-nspawn's --network-veth switch now gained a short
9843 version "-n", since with the changes above it is now truly
9844 useful out-of-the-box. The systemd-nspawn@.service has been
9845 updated to make use of it too by default.
9847 * systemd-nspawn will now maintain a per-image R/W lock, to
9848 ensure that the same image is not started more than once
9849 writable. (It's OK to run an image multiple times
9850 simultaneously in read-only mode.)
9852 * systemd-nspawn's --image= option is now capable of
9853 dissecting and booting MBR and GPT disk images that contain
9854 only a single active Linux partition. Previously it
9855 supported only GPT disk images with proper GPT type
9856 IDs. This allows running cloud images from major
9857 distributions directly with systemd-nspawn, without
9860 * In addition to collecting mouse dpi data in the udev
9861 hardware database, there's now support for collecting angle
9862 information for mouse scroll wheels. The database is
9863 supposed to guarantee similar scrolling behavior on mice
9864 that it knows about. There's also support for collecting
9865 information about Touchpad types.
9867 * udev's input_id built-in will now also collect touch screen
9868 dimension data and attach it to probed devices.
9870 * /etc/os-release gained support for a Distribution Privacy
9873 * networkd gained support for creating "ipvlan", "gretap",
9874 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
9876 * systemd-tmpfiles gained support for "a" lines for setting
9879 * systemd-nspawn will now mount /tmp in the container to
9880 tmpfs, automatically.
9882 * systemd now exposes the memory.usage_in_bytes cgroup
9883 attribute and shows it for each service in the "systemctl
9884 status" output, if available.
9886 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
9887 immediate reboot is triggered. This useful if shutdown is
9888 hung and is unable to complete, to expedite the
9889 operation. Note that this kind of reboot will still unmount
9890 all file systems, and hence should not result in fsck being
9893 * A .device unit for an optical block device will now be
9894 considered active only when a medium is in the drive. Also,
9895 mount units are now bound to their backing devices thus
9896 triggering automatic unmounting when devices become
9897 unavailable. With this in place systemd will now
9898 automatically unmount left-over mounts when a CD-ROM is
9899 ejected or an USB stick is yanked from the system.
9901 * networkd-wait-online now has support for waiting for
9902 specific interfaces only (with globbing), and for giving up
9903 after a configurable timeout.
9905 * networkd now exits when idle. It will be automatically
9906 restarted as soon as interfaces show up, are removed or
9907 change state. networkd will stay around as long as there is
9908 at least one DHCP state machine or similar around, that keep
9911 * networkd may now configure IPv6 link-local addressing in
9912 addition to IPv4 link-local addressing.
9914 * The IPv6 "token" for use in SLAAC may now be configured for
9915 each .network interface in networkd.
9917 * Routes configured with networkd may now be assigned a scope
9920 * networkd's [Match] sections now support globbing and lists
9921 of multiple space-separated matches per item.
9923 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
9924 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
9925 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
9926 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
9927 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
9928 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
9929 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
9930 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
9931 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
9932 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
9933 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
9934 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
9935 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
9936 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
9937 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
9938 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
9939 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
9940 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
9941 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
9942 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
9943 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
9944 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
9945 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
9946 Hoffmann, Zbigniew Jędrzejewski-Szmek
9948 — Berlin, 2015-02-16
9952 * When querying unit file enablement status (for example via
9953 "systemctl is-enabled"), a new state "indirect" is now known
9954 which indicates that a unit might not be enabled itself, but
9955 another unit listed in its Also= setting might be.
9957 * Similar to the various existing ConditionXYZ= settings for
9958 units, there are now matching AssertXYZ= settings. While
9959 failing conditions cause a unit to be skipped, but its job
9960 to succeed, failing assertions declared like this will cause
9961 a unit start operation and its job to fail.
9963 * hostnamed now knows a new chassis type "embedded".
9965 * systemctl gained a new "edit" command. When used on a unit
9966 file, this allows extending unit files with .d/ drop-in
9967 configuration snippets or editing the full file (after
9968 copying it from /usr/lib to /etc). This will invoke the
9969 user's editor (as configured with $EDITOR), and reload the
9970 modified configuration after editing.
9972 * "systemctl status" now shows the suggested enablement state
9973 for a unit, as declared in the (usually vendor-supplied)
9974 system preset files.
9976 * nss-myhostname will now resolve the single-label hostname
9977 "gateway" to the locally configured default IP routing
9978 gateways, ordered by their metrics. This assigns a stable
9979 name to the used gateways, regardless which ones are
9980 currently configured. Note that the name will only be
9981 resolved after all other name sources (if nss-myhostname is
9982 configured properly) and should hence not negatively impact
9983 systems that use the single-label hostname "gateway" in
9986 * systemd-inhibit now allows filtering by mode when listing
9989 * Scope and service units gained a new "Delegate" boolean
9990 property, which, when set, allows processes running inside the
9991 unit to further partition resources. This is primarily
9992 useful for systemd user instances as well as container
9995 * journald will now pick up audit messages directly from
9996 the kernel, and log them like any other log message. The
9997 audit fields are split up and fully indexed. This means that
9998 journalctl in many ways is now a (nicer!) alternative to
9999 ausearch, the traditional audit client. Note that this
10000 implements only a minimal audit client. If you want the
10001 special audit modes like reboot-on-log-overflow, please use
10002 the traditional auditd instead, which can be used in
10003 parallel to journald.
10005 * The ConditionSecurity= unit file option now understands the
10006 special string "audit" to check whether auditing is
10009 * journalctl gained two new commands --vacuum-size= and
10010 --vacuum-time= to delete old journal files until the
10011 remaining ones take up no more than the specified size on disk,
10012 or are not older than the specified time.
10014 * A new, native PPPoE library has been added to sd-network,
10015 systemd's library of light-weight networking protocols. This
10016 library will be used in a future version of networkd to
10017 enable PPPoE communication without an external pppd daemon.
10019 * The busctl tool now understands a new "capture" verb that
10020 works similar to "monitor", but writes a packet capture
10021 trace to STDOUT that can be redirected to a file which is
10022 compatible with libcap's capture file format. This can then
10023 be loaded in Wireshark and similar tools to inspect bus
10026 * The busctl tool now understands a new "tree" verb that shows
10027 the object trees of a specific service on the bus, or of all
10030 * The busctl tool now understands a new "introspect" verb that
10031 shows all interfaces and members of objects on the bus,
10032 including their signature and values. This is particularly
10033 useful to get more information about bus objects shown by
10034 the new "busctl tree" command.
10036 * The busctl tool now understands new verbs "call",
10037 "set-property" and "get-property" for invoking bus method
10038 calls, setting and getting bus object properties in a
10041 * busctl gained a new --augment-creds= argument that controls
10042 whether the tool shall augment credential information it
10043 gets from the bus with data from /proc, in a possibly
10046 * nspawn's --link-journal= switch gained two new values
10047 "try-guest" and "try-host" that work like "guest" and
10048 "host", but do not fail if the host has no persistent
10049 journaling enabled. -j is now equivalent to
10050 --link-journal=try-guest.
10052 * macvlan network devices created by nspawn will now have
10053 stable MAC addresses.
10055 * A new SmackProcessLabel= unit setting has been added, which
10056 controls the SMACK security label processes forked off by
10057 the respective unit shall use.
10059 * If compiled with --enable-xkbcommon, systemd-localed will
10060 verify x11 keymap settings by compiling the given keymap. It
10061 will spew out warnings if the compilation fails. This
10062 requires libxkbcommon to be installed.
10064 * When a coredump is collected, a larger number of metadata
10065 fields is now collected and included in the journal records
10066 created for it. More specifically, control group membership,
10067 environment variables, memory maps, working directory,
10068 chroot directory, /proc/$PID/status, and a list of open file
10069 descriptors is now stored in the log entry.
10071 * The udev hwdb now contains DPI information for mice. For
10074 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
10076 * All systemd programs that read standalone configuration
10077 files in /etc now also support a corresponding series of
10078 .conf.d configuration directories in /etc/, /run/,
10079 /usr/local/lib/, /usr/lib/, and (if configured with
10080 --enable-split-usr) /lib/. In particular, the following
10081 configuration files now have corresponding configuration
10082 directories: system.conf user.conf, logind.conf,
10083 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
10084 resolved.conf, timesyncd.conf, journal-remote.conf, and
10085 journal-upload.conf. Note that distributions should use the
10086 configuration directories in /usr/lib/; the directories in
10087 /etc/ are reserved for the system administrator.
10089 * systemd-rfkill will no longer take the rfkill device name
10090 into account when storing rfkill state on disk, as the name
10091 might be dynamically assigned and not stable. Instead, the
10092 ID_PATH udev variable combined with the rfkill type (wlan,
10093 bluetooth, …) is used.
10095 * A new service systemd-machine-id-commit.service has been
10096 added. When used on systems where /etc is read-only during
10097 boot, and /etc/machine-id is not initialized (but an empty
10098 file), this service will copy the temporary machine ID
10099 created as replacement into /etc after the system is fully
10100 booted up. This is useful for systems that are freshly
10101 installed with a non-initialized machine ID, but should get
10102 a fixed machine ID for subsequent boots.
10104 * networkd's .netdev files now provide a large set of
10105 configuration parameters for VXLAN devices. Similarly, the
10106 bridge port cost parameter is now configurable in .network
10107 files. There's also new support for configuring IP source
10108 routing. networkd .link files gained support for a new
10109 OriginalName= match that is useful to match against the
10110 original interface name the kernel assigned. .network files
10111 may include MTU= and MACAddress= fields for altering the MTU
10112 and MAC address while being connected to a specific network
10115 * The LUKS logic gained supported for configuring
10116 UUID-specific key files. There's also new support for naming
10117 LUKS device from the kernel command line, using the new
10118 luks.name= argument.
10120 * Timer units may now be transiently created via the bus API
10121 (this was previously already available for scope and service
10122 units). In addition it is now possible to create multiple
10123 transient units at the same time with a single bus call. The
10124 "systemd-run" tool has been updated to make use of this for
10125 running commands on a specified time, in at(1)-style.
10127 * tmpfiles gained support for "t" lines, for assigning
10128 extended attributes to files. Among other uses this may be
10129 used to assign SMACK labels to files.
10131 Contributions from: Alin Rauta, Alison Chaiken, Andrej
10132 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
10133 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
10134 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
10135 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
10136 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
10137 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
10138 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
10139 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
10140 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
10141 Michael Biebl, Michael Chapman, Michael Marineau, Michal
10142 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
10143 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
10144 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
10145 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
10146 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10147 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
10148 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
10150 — Berlin, 2014-12-10
10154 * journalctl gained the new options -t/--identifier= to match
10155 on the syslog identifier (aka "tag"), as well as --utc to
10156 show log timestamps in the UTC timezone. journalctl now also
10157 accepts -n/--lines=all to disable line capping in a pager.
10159 * journalctl gained a new switch, --flush, that synchronously
10160 flushes logs from /run/log/journal to /var/log/journal if
10161 persistent storage is enabled. systemd-journal-flush.service
10162 now waits until the operation is complete.
10164 * Services can notify the manager before they start a reload
10165 (by sending RELOADING=1) or shutdown (by sending
10166 STOPPING=1). This allows the manager to track and show the
10167 internal state of daemons and closes a race condition when
10168 the process is still running but has closed its D-Bus
10171 * Services with Type=oneshot do not have to have any ExecStart
10174 * User units are now loaded also from
10175 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
10176 /run/systemd/user directory that was already previously
10177 supported, but is under the control of the user.
10179 * Job timeouts (i.e. timeouts on the time a job that is
10180 queued stays in the run queue) can now optionally result in
10181 immediate reboot or power-off actions (JobTimeoutAction= and
10182 JobTimeoutRebootArgument=). This is useful on ".target"
10183 units, to limit the maximum time a target remains
10184 undispatched in the run queue, and to trigger an emergency
10185 operation in such a case. This is now used by default to
10186 turn off the system if boot-up (as defined by everything in
10187 basic.target) hangs and does not complete for at least
10188 15min. Also, if power-off or reboot hang for at least 30min
10189 an immediate power-off/reboot operation is triggered. This
10190 functionality is particularly useful to increase reliability
10191 on embedded devices, but also on laptops which might
10192 accidentally get powered on when carried in a backpack and
10193 whose boot stays stuck in a hard disk encryption passphrase
10196 * systemd-logind can be configured to also handle lid switch
10197 events even when the machine is docked or multiple displays
10198 are attached (HandleLidSwitchDocked= option).
10200 * A helper binary and a service have been added which can be
10201 used to resume from hibernation in the initramfs. A
10202 generator will parse the resume= option on the kernel
10203 command line to trigger resume.
10205 * A user console daemon systemd-consoled has been
10206 added. Currently, it is a preview, and will so far open a
10207 single terminal on each session of the user marked as
10208 Desktop=systemd-console.
10210 * Route metrics can be specified for DHCP routes added by
10213 * The SELinux context of socket-activated services can be set
10214 from the information provided by the networking stack
10215 (SELinuxContextFromNet= option).
10217 * Userspace firmware loading support has been removed and
10218 the minimum supported kernel version is thus bumped to 3.7.
10220 * Timeout for udev workers has been increased from 1 to 3
10221 minutes, but a warning will be printed after 1 minute to
10222 help diagnose kernel modules that take a long time to load.
10224 * Udev rules can now remove tags on devices with TAG-="foobar".
10226 * systemd's readahead implementation has been removed. In many
10227 circumstances it didn't give expected benefits even for
10228 rotational disk drives and was becoming less relevant in the
10229 age of SSDs. As none of the developers has been using
10230 rotating media anymore, and nobody stepped up to actively
10231 maintain this component of systemd it has now been removed.
10233 * Swap units can use Options= to specify discard options.
10234 Discard options specified for swaps in /etc/fstab are now
10237 * Docker containers are now detected as a separate type of
10240 * The Password Agent protocol gained support for queries where
10241 the user input is shown, useful e.g. for user names.
10242 systemd-ask-password gained a new --echo option to turn that
10245 * The default sysctl.d/ snippets will now set:
10247 net.core.default_qdisc = fq_codel
10249 This selects Fair Queuing Controlled Delay as the default
10250 queuing discipline for network interfaces. fq_codel helps
10251 fight the network bufferbloat problem. It is believed to be
10252 a good default with no tuning required for most workloads.
10253 Downstream distributions may override this choice. On 10Gbit
10254 servers that do not do forwarding, "fq" may perform better.
10255 Systems without a good clocksource should use "pfifo_fast".
10257 * If kdbus is enabled during build a new option BusPolicy= is
10258 available for service units, that allows locking all service
10259 processes into a stricter bus policy, in order to limit
10260 access to various bus services, or even hide most of them
10261 from the service's view entirely.
10263 * networkctl will now show the .network and .link file
10264 networkd has applied to a specific interface.
10266 * sd-login gained a new API call sd_session_get_desktop() to
10267 query which desktop environment has been selected for a
10270 * UNIX utmp support is now compile-time optional to support
10271 legacy-free systems.
10273 * systemctl gained two new commands "add-wants" and
10274 "add-requires" for pulling in units from specific targets
10277 * If the word "rescue" is specified on the kernel command line
10278 the system will now boot into rescue mode (aka
10279 rescue.target), which was previously available only by
10280 specifying "1" or "systemd.unit=rescue.target" on the kernel
10281 command line. This new kernel command line option nicely
10282 mirrors the already existing "emergency" kernel command line
10285 * New kernel command line options mount.usr=, mount.usrflags=,
10286 mount.usrfstype= have been added that match root=, rootflags=,
10287 rootfstype= but allow mounting a specific file system to
10290 * The $NOTIFY_SOCKET is now also passed to control processes of
10291 services, not only the main process.
10293 * This version reenables support for fsck's -l switch. This
10294 means at least version v2.25 of util-linux is required for
10295 operation, otherwise dead-locks on device nodes may
10296 occur. Again: you need to update util-linux to at least
10297 v2.25 when updating systemd to v217.
10299 * The "multi-seat-x" tool has been removed from systemd, as
10300 its functionality has been integrated into X servers 1.16,
10301 and the tool is hence redundant. It is recommended to update
10302 display managers invoking this tool to simply invoke X
10303 directly from now on, again.
10305 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
10306 message flag has been added for all of systemd's polkit
10307 authenticated method calls has been added. In particular this
10308 now allows optional interactive authorization via polkit for
10309 many of PID1's privileged operations such as unit file
10310 enabling and disabling.
10312 * "udevadm hwdb --update" learnt a new switch "--usr" for
10313 placing the rebuilt hardware database in /usr instead of
10314 /etc. When used only hardware database entries stored in
10315 /usr will be used, and any user database entries in /etc are
10316 ignored. This functionality is useful for vendors to ship a
10317 pre-built database on systems where local configuration is
10318 unnecessary or unlikely.
10320 * Calendar time specifications in .timer units now also
10321 understand the strings "semi-annually", "quarterly" and
10322 "minutely" as shortcuts (in addition to the preexisting
10323 "annually", "hourly", …).
10325 * systemd-tmpfiles will now correctly create files in /dev
10326 at boot which are marked for creation only at boot. It is
10327 recommended to always create static device nodes with 'c!'
10328 and 'b!', so that they are created only at boot and not
10329 overwritten at runtime.
10331 * When the watchdog logic is used for a service (WatchdogSec=)
10332 and the watchdog timeout is hit the service will now be
10333 terminated with SIGABRT (instead of just SIGTERM), in order
10334 to make sure a proper coredump and backtrace is
10335 generated. This ensures that hanging services will result in
10336 similar coredump/backtrace behaviour as services that hit a
10337 segmentation fault.
10339 Contributions from: Andreas Henriksson, Andrei Borzenkov,
10340 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
10341 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
10342 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
10343 Herrmann, David Sommerseth, David Strauss, Emil Renner
10344 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
10345 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
10346 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
10347 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
10348 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
10349 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
10350 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
10351 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
10352 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
10353 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
10354 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
10355 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
10356 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
10357 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
10358 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
10359 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
10362 — Berlin, 2014-10-28
10366 * timedated no longer reads NTP implementation unit names from
10367 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
10368 implementations should add a
10370 Conflicts=systemd-timesyncd.service
10372 to their unit files to take over and replace systemd's NTP
10373 default functionality.
10375 * systemd-sysusers gained a new line type "r" for configuring
10376 which UID/GID ranges to allocate system users/groups
10377 from. Lines of type "u" may now add an additional column
10378 that specifies the home directory for the system user to be
10379 created. Also, systemd-sysusers may now optionally read user
10380 information from STDIN instead of a file. This is useful for
10381 invoking it from RPM preinst scriptlets that need to create
10382 users before the first RPM file is installed since these
10383 files might need to be owned by them. A new
10384 %sysusers_create_inline RPM macro has been introduced to do
10385 just that. systemd-sysusers now updates the shadow files as
10386 well as the user/group databases, which should enhance
10387 compatibility with certain tools like grpck.
10389 * A number of bus APIs of PID 1 now optionally consult polkit to
10390 permit access for otherwise unprivileged clients under certain
10391 conditions. Note that this currently doesn't support
10392 interactive authentication yet, but this is expected to be
10393 added eventually, too.
10395 * /etc/machine-info now has new fields for configuring the
10396 deployment environment of the machine, as well as the
10397 location of the machine. hostnamectl has been updated with
10398 new command to update these fields.
10400 * systemd-timesyncd has been updated to automatically acquire
10401 NTP server information from systemd-networkd, which might
10402 have been discovered via DHCP.
10404 * systemd-resolved now includes a caching DNS stub resolver
10405 and a complete LLMNR name resolution implementation. A new
10406 NSS module "nss-resolve" has been added which can be used
10407 instead of glibc's own "nss-dns" to resolve hostnames via
10408 systemd-resolved. Hostnames, addresses and arbitrary RRs may
10409 be resolved via systemd-resolved D-Bus APIs. In contrast to
10410 the glibc internal resolver systemd-resolved is aware of
10411 multi-homed system, and keeps DNS server and caches separate
10412 and per-interface. Queries are sent simultaneously on all
10413 interfaces that have DNS servers configured, in order to
10414 properly handle VPNs and local LANs which might resolve
10415 separate sets of domain names. systemd-resolved may acquire
10416 DNS server information from systemd-networkd automatically,
10417 which in turn might have discovered them via DHCP. A tool
10418 "systemd-resolve-host" has been added that may be used to
10419 query the DNS logic in resolved. systemd-resolved implements
10420 IDNA and automatically uses IDNA or UTF-8 encoding depending
10421 on whether classic DNS or LLMNR is used as transport. In the
10422 next releases we intend to add a DNSSEC and mDNS/DNS-SD
10423 implementation to systemd-resolved.
10425 * A new NSS module nss-mymachines has been added, that
10426 automatically resolves the names of all local registered
10427 containers to their respective IP addresses.
10429 * A new client tool "networkctl" for systemd-networkd has been
10430 added. It currently is entirely passive and will query
10431 networking configuration from udev, rtnetlink and networkd,
10432 and present it to the user in a very friendly
10433 way. Eventually, we hope to extend it to become a full
10434 control utility for networkd.
10436 * .socket units gained a new DeferAcceptSec= setting that
10437 controls the kernels' TCP_DEFER_ACCEPT sockopt for
10438 TCP. Similarly, support for controlling TCP keep-alive
10439 settings has been added (KeepAliveTimeSec=,
10440 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
10441 turning off Nagle's algorithm on TCP has been added
10444 * logind learned a new session type "web", for use in projects
10445 like Cockpit which register web clients as PAM sessions.
10447 * timer units with at least one OnCalendar= setting will now
10448 be started only after time-sync.target has been
10449 reached. This way they will not elapse before the system
10450 clock has been corrected by a local NTP client or
10451 similar. This is particular useful on RTC-less embedded
10452 machines, that come up with an invalid system clock.
10454 * systemd-nspawn's --network-veth= switch should now result in
10455 stable MAC addresses for both the outer and the inner side
10458 * systemd-nspawn gained a new --volatile= switch for running
10459 container instances with /etc or /var unpopulated.
10461 * The kdbus client code has been updated to use the new Linux
10462 3.17 memfd subsystem instead of the old kdbus-specific one.
10464 * systemd-networkd's DHCP client and server now support
10465 FORCERENEW. There are also new configuration options to
10466 configure the vendor client identifier and broadcast mode
10469 * systemd will no longer inform the kernel about the current
10470 timezone, as this is necessarily incorrect and racy as the
10471 kernel has no understanding of DST and similar
10472 concepts. This hence means FAT timestamps will be always
10473 considered UTC, similar to what Android is already
10474 doing. Also, when the RTC is configured to the local time
10475 (rather than UTC) systemd will never synchronize back to it,
10476 as this might confuse Windows at a later boot.
10478 * systemd-analyze gained a new command "verify" for offline
10479 validation of unit files.
10481 * systemd-networkd gained support for a couple of additional
10482 settings for bonding networking setups. Also, the metric for
10483 statically configured routes may now be configured. For
10484 network interfaces where this is appropriate the peer IP
10485 address may now be configured.
10487 * systemd-networkd's DHCP client will no longer request
10488 broadcasting by default, as this tripped up some networks.
10489 For hardware where broadcast is required the feature should
10490 be switched back on using RequestBroadcast=yes.
10492 * systemd-networkd will now set up IPv4LL addresses (when
10493 enabled) even if DHCP is configured successfully.
10495 * udev will now default to respect network device names given
10496 by the kernel when the kernel indicates that these are
10497 predictable. This behavior can be tweaked by changing
10498 NamePolicy= in the relevant .link file.
10500 * A new library systemd-terminal has been added that
10501 implements full TTY stream parsing and rendering. This
10502 library is supposed to be used later on for implementing a
10503 full userspace VT subsystem, replacing the current kernel
10506 * A new tool systemd-journal-upload has been added to push
10507 journal data to a remote system running
10508 systemd-journal-remote.
10510 * journald will no longer forward all local data to another
10511 running syslog daemon. This change has been made because
10512 rsyslog (which appears to be the most commonly used syslog
10513 implementation these days) no longer makes use of this, and
10514 instead pulls the data out of the journal on its own. Since
10515 forwarding the messages to a non-existent syslog server is
10516 more expensive than we assumed we have now turned this
10517 off. If you run a syslog server that is not a recent rsyslog
10518 version, you have to turn this option on again
10519 (ForwardToSyslog= in journald.conf).
10521 * journald now optionally supports the LZ4 compressor for
10522 larger journal fields. This compressor should perform much
10523 better than XZ which was the previous default.
10525 * machinectl now shows the IP addresses of local containers,
10526 if it knows them, plus the interface name of the container.
10528 * A new tool "systemd-escape" has been added that makes it
10529 easy to escape strings to build unit names and similar.
10531 * sd_notify() messages may now include a new ERRNO= field
10532 which is parsed and collected by systemd and shown among the
10533 "systemctl status" output for a service.
10535 * A new component "systemd-firstboot" has been added that
10536 queries the most basic systemd information (timezone,
10537 hostname, root password) interactively on first
10538 boot. Alternatively it may also be used to provision these
10539 things offline on OS images installed into directories.
10541 * The default sysctl.d/ snippets will now set
10543 net.ipv4.conf.default.promote_secondaries=1
10545 This has the benefit of no flushing secondary IP addresses
10546 when primary addresses are removed.
10548 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
10549 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
10550 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
10551 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
10552 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
10553 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
10554 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
10555 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
10556 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
10557 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
10558 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
10559 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
10560 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
10561 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
10562 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
10564 — Berlin, 2014-08-19
10568 * A new tool systemd-sysusers has been added. This tool
10569 creates system users and groups in /etc/passwd and
10570 /etc/group, based on static declarative system user/group
10571 definitions in /usr/lib/sysusers.d/. This is useful to
10572 enable factory resets and volatile systems that boot up with
10573 an empty /etc directory, and thus need system users and
10574 groups created during early boot. systemd now also ships
10575 with two default sysusers.d/ files for the most basic
10576 users and groups systemd and the core operating system
10579 * A new tmpfiles snippet has been added that rebuilds the
10580 essential files in /etc on boot, should they be missing.
10582 * A directive for ensuring automatic clean-up of
10583 /var/cache/man/ has been removed from the default
10584 configuration. This line should now be shipped by the man
10585 implementation. The necessary change has been made to the
10586 man-db implementation. Note that you need to update your man
10587 implementation to one that ships this line, otherwise no
10588 automatic clean-up of /var/cache/man will take place.
10590 * A new condition ConditionNeedsUpdate= has been added that
10591 may conditionalize services to only run when /etc or /var
10592 are "older" than the vendor operating system resources in
10593 /usr. This is useful for reconstructing or updating /etc
10594 after an offline update of /usr or a factory reset, on the
10595 next reboot. Services that want to run once after such an
10596 update or reset should use this condition and order
10597 themselves before the new systemd-update-done.service, which
10598 will mark the two directories as fully updated. A number of
10599 service files have been added making use of this, to rebuild
10600 the udev hardware database, the journald message catalog and
10601 dynamic loader cache (ldconfig). The systemd-sysusers tool
10602 described above also makes use of this now. With this in
10603 place it is now possible to start up a minimal operating
10604 system with /etc empty cleanly. For more information on the
10605 concepts involved see this recent blog story:
10607 http://0pointer.de/blog/projects/stateless.html
10609 * A new system group "input" has been introduced, and all
10610 input device nodes get this group assigned. This is useful
10611 for system-level software to get access to input devices. It
10612 complements what is already done for "audio" and "video".
10614 * systemd-networkd learnt minimal DHCPv4 server support in
10615 addition to the existing DHCPv4 client support. It also
10616 learnt DHCPv6 client and IPv6 Router Solicitation client
10617 support. The DHCPv4 client gained support for static routes
10618 passed in from the server. Note that the [DHCPv4] section
10619 known in older systemd-networkd versions has been renamed to
10620 [DHCP] and is now also used by the DHCPv6 client. Existing
10621 .network files using settings of this section should be
10622 updated, though compatibility is maintained. Optionally, the
10623 client hostname may now be sent to the DHCP server.
10625 * networkd gained support for vxlan virtual networks as well
10626 as tun/tap and dummy devices.
10628 * networkd gained support for automatic allocation of address
10629 ranges for interfaces from a system-wide pool of
10630 addresses. This is useful for dynamically managing a large
10631 number of interfaces with a single network configuration
10632 file. In particular this is useful to easily assign
10633 appropriate IP addresses to the veth links of a large number
10634 of nspawn instances.
10636 * RPM macros for processing sysusers, sysctl and binfmt
10637 drop-in snippets at package installation time have been
10640 * The /etc/os-release file should now be placed in
10641 /usr/lib/os-release. The old location is automatically
10642 created as symlink. /usr/lib is the more appropriate
10643 location of this file, since it shall actually describe the
10644 vendor operating system shipped in /usr, and not the
10645 configuration stored in /etc.
10647 * .mount units gained a new boolean SloppyOptions= setting
10648 that maps to mount(8)'s -s option which enables permissive
10649 parsing of unknown mount options.
10651 * tmpfiles learnt a new "L+" directive which creates a symlink
10652 but (unlike "L") deletes a pre-existing file first, should
10653 it already exist and not already be the correct
10654 symlink. Similarly, "b+", "c+" and "p+" directives have been
10655 added as well, which create block and character devices, as
10656 well as fifos in the filesystem, possibly removing any
10657 pre-existing files of different types.
10659 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
10660 'argument' field (which so far specified the source to
10661 symlink/copy the files from) is now optional. If omitted the
10662 same file os copied from /usr/share/factory/ suffixed by the
10663 full destination path. This is useful for populating /etc
10664 with essential files, by copying them from vendor defaults
10665 shipped in /usr/share/factory/etc.
10667 * A new command "systemctl preset-all" has been added that
10668 applies the service preset settings to all installed unit
10669 files. A new switch --preset-mode= has been added that
10670 controls whether only enable or only disable operations
10673 * A new command "systemctl is-system-running" has been added
10674 that allows checking the overall state of the system, for
10675 example whether it is fully up and running.
10677 * When the system boots up with an empty /etc, the equivalent
10678 to "systemctl preset-all" is executed during early boot, to
10679 make sure all default services are enabled after a factory
10682 * systemd now contains a minimal preset file that enables the
10683 most basic services systemd ships by default.
10685 * Unit files' [Install] section gained a new DefaultInstance=
10686 field for defining the default instance to create if a
10687 template unit is enabled with no instance specified.
10689 * A new passive target cryptsetup-pre.target has been added
10690 that may be used by services that need to make they run and
10691 finish before the first LUKS cryptographic device is set up.
10693 * The /dev/loop-control and /dev/btrfs-control device nodes
10694 are now owned by the "disk" group by default, opening up
10695 access to this group.
10697 * systemd-coredump will now automatically generate a
10698 stack trace of all core dumps taking place on the system,
10699 based on elfutils' libdw library. This stack trace is logged
10702 * systemd-coredump may now optionally store coredumps directly
10703 on disk (in /var/lib/systemd/coredump, possibly compressed),
10704 instead of storing them unconditionally in the journal. This
10705 mode is the new default. A new configuration file
10706 /etc/systemd/coredump.conf has been added to configure this
10707 and other parameters of systemd-coredump.
10709 * coredumpctl gained a new "info" verb to show details about a
10710 specific coredump. A new switch "-1" has also been added
10711 that makes sure to only show information about the most
10712 recent entry instead of all entries. Also, as the tool is
10713 generally useful now the "systemd-" prefix of the binary
10714 name has been removed. Distributions that want to maintain
10715 compatibility with the old name should add a symlink from
10716 the old name to the new name.
10718 * journald's SplitMode= now defaults to "uid". This makes sure
10719 that unprivileged users can access their own coredumps with
10720 coredumpctl without restrictions.
10722 * New kernel command line options "systemd.wants=" (for
10723 pulling an additional unit during boot), "systemd.mask="
10724 (for masking a specific unit for the boot), and
10725 "systemd.debug-shell" (for enabling the debug shell on tty9)
10726 have been added. This is implemented in the new generator
10727 "systemd-debug-generator".
10729 * systemd-nspawn will now by default filter a couple of
10730 syscalls for containers, among them those required for
10731 kernel module loading, direct x86 IO port access, swap
10732 management, and kexec. Most importantly though
10733 open_by_handle_at() is now prohibited for containers,
10734 closing a hole similar to a recently discussed vulnerability
10735 in docker regarding access to files on file hierarchies the
10736 container should normally not have access to. Note that, for
10737 nspawn, we generally make no security claims anyway (and
10738 this is explicitly documented in the man page), so this is
10739 just a fix for one of the most obvious problems.
10741 * A new man page file-hierarchy(7) has been added that
10742 contains a minimized, modernized version of the file system
10743 layout systemd expects, similar in style to the FHS
10744 specification or hier(5). A new tool systemd-path(1) has
10745 been added to query many of these paths for the local
10748 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
10749 longer done. Since the directory now has a per-user size
10750 limit, and is cleaned on logout this appears unnecessary,
10751 in particular since this now brings the lifecycle of this
10752 directory closer in line with how IPC objects are handled.
10754 * systemd.pc now exports a number of additional directories,
10755 including $libdir (which is useful to identify the library
10756 path for the primary architecture of the system), and a
10757 couple of drop-in directories.
10759 * udev's predictable network interface names now use the dev_port
10760 sysfs attribute, introduced in linux 3.15 instead of dev_id to
10761 distinguish between ports of the same PCI function. dev_id should
10762 only be used for ports using the same HW address, hence the need
10765 * machined has been updated to export the OS version of a
10766 container (read from /etc/os-release and
10767 /usr/lib/os-release) on the bus. This is now shown in
10768 "machinectl status" for a machine.
10770 * A new service setting RestartForceExitStatus= has been
10771 added. If configured to a set of exit signals or process
10772 return values, the service will be restarted when the main
10773 daemon process exits with any of them, regardless of the
10776 * systemctl's -H switch for connecting to remote systemd
10777 machines has been extended so that it may be used to
10778 directly connect to a specific container on the
10779 host. "systemctl -H root@foobar:waldi" will now connect as
10780 user "root" to host "foobar", and then proceed directly to
10781 the container named "waldi". Note that currently you have to
10782 authenticate as user "root" for this to work, as entering
10783 containers is a privileged operation.
10785 Contributions from: Andreas Henriksson, Benjamin Steinwender,
10786 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
10787 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
10788 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
10789 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
10790 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
10791 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
10792 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
10793 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
10794 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
10795 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
10796 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
10798 — Berlin, 2014-07-03
10802 * As an experimental feature, udev now tries to lock the
10803 disk device node (flock(LOCK_SH|LOCK_NB)) while it
10804 executes events for the disk or any of its partitions.
10805 Applications like partitioning programs can lock the
10806 disk device node (flock(LOCK_EX)) and claim temporary
10807 device ownership that way; udev will entirely skip all event
10808 handling for this disk and its partitions. If the disk
10809 was opened for writing, the close will trigger a partition
10810 table rescan in udev's "watch" facility, and if needed
10811 synthesize "change" events for the disk and all its partitions.
10812 This is now unconditionally enabled, and if it turns out to
10813 cause major problems, we might turn it on only for specific
10814 devices, or might need to disable it entirely. Device Mapper
10815 devices are excluded from this logic.
10817 * We temporarily dropped the "-l" switch for fsck invocations,
10818 since they collide with the flock() logic above. util-linux
10819 upstream has been changed already to avoid this conflict,
10820 and we will re-add "-l" as soon as util-linux with this
10821 change has been released.
10823 * The dependency on libattr has been removed. Since a long
10824 time, the extended attribute calls have moved to glibc, and
10825 libattr is thus unnecessary.
10827 * Virtualization detection works without privileges now. This
10828 means the systemd-detect-virt binary no longer requires
10829 CAP_SYS_PTRACE file capabilities, and our daemons can run
10830 with fewer privileges.
10832 * systemd-networkd now runs under its own "systemd-network"
10833 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
10834 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
10835 loses the ability to write to files owned by root this way.
10837 * Similarly, systemd-resolved now runs under its own
10838 "systemd-resolve" user with no capabilities remaining.
10840 * Similarly, systemd-bus-proxyd now runs under its own
10841 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
10843 * systemd-networkd gained support for setting up "veth"
10844 virtual Ethernet devices for container connectivity, as well
10845 as GRE and VTI tunnels.
10847 * systemd-networkd will no longer automatically attempt to
10848 manually load kernel modules necessary for certain tunnel
10849 transports. Instead, it is assumed the kernel loads them
10850 automatically when required. This only works correctly on
10851 very new kernels. On older kernels, please consider adding
10852 the kernel modules to /etc/modules-load.d/ as a work-around.
10854 * The resolv.conf file systemd-resolved generates has been
10855 moved to /run/systemd/resolve/. If you have a symlink from
10856 /etc/resolv.conf, it might be necessary to correct it.
10858 * Two new service settings, ProtectHome= and ProtectSystem=,
10859 have been added. When enabled, they will make the user data
10860 (such as /home) inaccessible or read-only and the system
10861 (such as /usr) read-only, for specific services. This allows
10862 very light-weight per-service sandboxing to avoid
10863 modifications of user data or system files from
10864 services. These two new switches have been enabled for all
10865 of systemd's long-running services, where appropriate.
10867 * Socket units gained new SocketUser= and SocketGroup=
10868 settings to set the owner user and group of AF_UNIX sockets
10869 and FIFOs in the file system.
10871 * Socket units gained a new RemoveOnStop= setting. If enabled,
10872 all FIFOS and sockets in the file system will be removed
10873 when the specific socket unit is stopped.
10875 * Socket units gained a new Symlinks= setting. It takes a list
10876 of symlinks to create to file system sockets or FIFOs
10877 created by the specific Unix sockets. This is useful to
10878 manage symlinks to socket nodes with the same lifecycle as
10881 * The /dev/log socket and /dev/initctl FIFO have been moved to
10882 /run, and have been replaced by symlinks. This allows
10883 connecting to these facilities even if PrivateDevices=yes is
10884 used for a service (which makes /dev/log itself unavailable,
10885 but /run is left). This also has the benefit of ensuring
10886 that /dev only contains device nodes, directories and
10887 symlinks, and nothing else.
10889 * sd-daemon gained two new calls sd_pid_notify() and
10890 sd_pid_notifyf(). They are similar to sd_notify() and
10891 sd_notifyf(), but allow overriding of the source PID of
10892 notification messages if permissions permit this. This is
10893 useful to send notify messages on behalf of a different
10894 process (for example, the parent process). The
10895 systemd-notify tool has been updated to make use of this
10896 when sending messages (so that notification messages now
10897 originate from the shell script invoking systemd-notify and
10898 not the systemd-notify process itself. This should minimize
10899 a race where systemd fails to associate notification
10900 messages to services when the originating process already
10903 * A new "on-abnormal" setting for Restart= has been added. If
10904 set, it will result in automatic restarts on all "abnormal"
10905 reasons for a process to exit, which includes unclean
10906 signals, core dumps, timeouts and watchdog timeouts, but
10907 does not include clean and unclean exit codes or clean
10908 signals. Restart=on-abnormal is an alternative for
10909 Restart=on-failure for services that shall be able to
10910 terminate and avoid restarts on certain errors, by
10911 indicating so with an unclean exit code. Restart=on-failure
10912 or Restart=on-abnormal is now the recommended setting for
10913 all long-running services.
10915 * If the InaccessibleDirectories= service setting points to a
10916 mount point (or if there are any submounts contained within
10917 it), it is now attempted to completely unmount it, to make
10918 the file systems truly unavailable for the respective
10921 * The ReadOnlyDirectories= service setting and
10922 systemd-nspawn's --read-only parameter are now recursively
10923 applied to all submounts, too.
10925 * Mount units may now be created transiently via the bus APIs.
10927 * The support for SysV and LSB init scripts has been removed
10928 from the systemd daemon itself. Instead, it is now
10929 implemented as a generator that creates native systemd units
10930 from these scripts when needed. This enables us to remove a
10931 substantial amount of legacy code from PID 1, following the
10932 fact that many distributions only ship a very small number
10933 of LSB/SysV init scripts nowadays.
10935 * Privileged Xen (dom0) domains are not considered
10936 virtualization anymore by the virtualization detection
10937 logic. After all, they generally have unrestricted access to
10938 the hardware and usually are used to manage the unprivileged
10941 * systemd-tmpfiles gained a new "C" line type, for copying
10942 files or entire directories.
10944 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
10945 lines. So far, they have been non-globbing versions of the
10946 latter, and have thus been redundant. In future, it is
10947 recommended to only use "z". "m" has hence been removed
10948 from the documentation, even though it stays supported.
10950 * A tmpfiles snippet to recreate the most basic structure in
10951 /var has been added. This is enough to create the /var/run →
10952 /run symlink and create a couple of structural
10953 directories. This allows systems to boot up with an empty or
10954 volatile /var. Of course, while with this change, the core OS
10955 now is capable with dealing with a volatile /var, not all
10956 user services are ready for it. However, we hope that sooner
10957 or later, many service daemons will be changed upstream so
10958 that they are able to automatically create their necessary
10959 directories in /var at boot, should they be missing. This is
10960 the first step to allow state-less systems that only require
10961 the vendor image for /usr to boot.
10963 * systemd-nspawn has gained a new --tmpfs= switch to mount an
10964 empty tmpfs instance to a specific directory. This is
10965 particularly useful for making use of the automatic
10966 reconstruction of /var (see above), by passing --tmpfs=/var.
10968 * Access modes specified in tmpfiles snippets may now be
10969 prefixed with "~", which indicates that they shall be masked
10970 by whether the existing file or directory is currently
10971 writable, readable or executable at all. Also, if specified,
10972 the sgid/suid/sticky bits will be masked for all
10975 * A new passive target unit "network-pre.target" has been
10976 added which is useful for services that shall run before any
10977 network is configured, for example firewall scripts.
10979 * The "floppy" group that previously owned the /dev/fd*
10980 devices is no longer used. The "disk" group is now used
10981 instead. Distributions should probably deprecate usage of
10984 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
10985 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
10986 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
10987 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
10988 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
10989 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
10992 — Berlin, 2014-06-11
10996 * A new "systemd-timesyncd" daemon has been added for
10997 synchronizing the system clock across the network. It
10998 implements an SNTP client. In contrast to NTP
10999 implementations such as chrony or the NTP reference server,
11000 this only implements a client side, and does not bother with
11001 the full NTP complexity, focusing only on querying time from
11002 one remote server and synchronizing the local clock to
11003 it. Unless you intend to serve NTP to networked clients or
11004 want to connect to local hardware clocks, this simple NTP
11005 client should be more than appropriate for most
11006 installations. The daemon runs with minimal privileges, and
11007 has been hooked up with networkd to only operate when
11008 network connectivity is available. The daemon saves the
11009 current clock to disk every time a new NTP sync has been
11010 acquired, and uses this to possibly correct the system clock
11011 early at bootup, in order to accommodate for systems that
11012 lack an RTC such as the Raspberry Pi and embedded devices,
11013 and to make sure that time monotonically progresses on these
11014 systems, even if it is not always correct. To make use of
11015 this daemon, a new system user and group "systemd-timesync"
11016 needs to be created on installation of systemd.
11018 * The queue "seqnum" interface of libudev has been disabled, as
11019 it was generally incompatible with device namespacing as
11020 sequence numbers of devices go "missing" if the devices are
11021 part of a different namespace.
11023 * "systemctl list-timers" and "systemctl list-sockets" gained
11024 a --recursive switch for showing units of these types also
11025 for all local containers, similar in style to the already
11026 supported --recursive switch for "systemctl list-units".
11028 * A new RebootArgument= setting has been added for service
11029 units, which may be used to specify a kernel reboot argument
11030 to use when triggering reboots with StartLimitAction=.
11032 * A new FailureAction= setting has been added for service
11033 units which may be used to specify an operation to trigger
11034 when a service fails. This works similarly to
11035 StartLimitAction=, but unlike it, controls what is done
11036 immediately rather than only after several attempts to
11037 restart the service in question.
11039 * hostnamed got updated to also expose the kernel name,
11040 release, and version on the bus. This is useful for
11041 executing commands like hostnamectl with the -H switch.
11042 systemd-analyze makes use of this to properly display
11043 details when running non-locally.
11045 * The bootchart tool can now show cgroup information in the
11046 graphs it generates.
11048 * The CFS CPU quota cgroup attribute is now exposed for
11049 services. The new CPUQuota= switch has been added for this
11050 which takes a percentage value. Setting this will have the
11051 result that a service may never get more CPU time than the
11052 specified percentage, even if the machine is otherwise idle.
11054 * systemd-networkd learned IPIP and SIT tunnel support.
11056 * LSB init scripts exposing a dependency on $network will now
11057 get a dependency on network-online.target rather than simply
11058 network.target. This should bring LSB handling closer to
11059 what it was on SysV systems.
11061 * A new fsck.repair= kernel option has been added to control
11062 how fsck shall deal with unclean file systems at boot.
11064 * The (.ini) configuration file parser will now silently ignore
11065 sections whose names begin with "X-". This may be used to maintain
11066 application-specific extension sections in unit files.
11068 * machined gained a new API to query the IP addresses of
11069 registered containers. "machinectl status" has been updated
11070 to show these addresses in its output.
11072 * A new call sd_uid_get_display() has been added to the
11073 sd-login APIs for querying the "primary" session of a
11074 user. The "primary" session of the user is elected from the
11075 user's sessions and generally a graphical session is
11076 preferred over a text one.
11078 * A minimal systemd-resolved daemon has been added. It
11079 currently simply acts as a companion to systemd-networkd and
11080 manages resolv.conf based on per-interface DNS
11081 configuration, possibly supplied via DHCP. In the long run
11082 we hope to extend this into a local DNSSEC enabled DNS and
11085 * The systemd-networkd-wait-online tool is now enabled by
11086 default. It will delay network-online.target until a network
11087 connection has been configured. The tool primarily integrates
11088 with networkd, but will also make a best effort to make sense
11089 of network configuration performed in some other way.
11091 * Two new service options StartupCPUShares= and
11092 StartupBlockIOWeight= have been added that work similarly to
11093 CPUShares= and BlockIOWeight= however only apply during
11094 system startup. This is useful to prioritize certain services
11095 differently during bootup than during normal runtime.
11097 * hostnamed has been changed to prefer the statically
11098 configured hostname in /etc/hostname (unless set to
11099 'localhost' or empty) over any dynamic one supplied by
11100 dhcp. With this change, the rules for picking the hostname
11101 match more closely the rules of other configuration settings
11102 where the local administrator's configuration in /etc always
11103 overrides any other settings.
11105 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
11106 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
11107 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
11108 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
11109 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
11110 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
11111 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
11112 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
11113 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
11114 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
11115 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
11116 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
11117 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
11118 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
11119 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
11120 Lindskog, WaLyong Cho, Will Woods, Zbigniew
11123 — Beijing, 2014-05-28
11127 * When restoring the screen brightness at boot, stay away from
11128 the darkest setting or from the lowest 5% of the available
11129 range, depending on which is the larger value of both. This
11130 should effectively protect the user from rebooting into a
11131 black screen, should the brightness have been set to minimum
11134 * sd-login gained a new sd_machine_get_class() call to
11135 determine the class ("vm" or "container") of a machine
11136 registered with machined.
11138 * sd-login gained new calls
11139 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
11140 to query the identity of the peer of a local AF_UNIX
11141 connection. They operate similarly to their sd_pid_get_xyz()
11144 * PID 1 will now maintain a system-wide system state engine
11145 with the states "starting", "running", "degraded",
11146 "maintenance", "stopping". These states are bound to system
11147 startup, normal runtime, runtime with at least one failed
11148 service, rescue/emergency mode and system shutdown. This
11149 state is shown in the "systemctl status" output when no unit
11150 name is passed. It is useful to determine system state, in
11151 particularly when doing so for many systems or containers at
11154 * A new command "list-machines" has been added to "systemctl"
11155 that lists all local OS containers and shows their system
11156 state (see above), if systemd runs inside of them.
11158 * systemctl gained a new "-r" switch to recursively enumerate
11159 units on all local containers, when used with the
11160 "list-unit" command (which is the default one that is
11161 executed when no parameters are specified).
11163 * The GPT automatic partition discovery logic will now honour
11164 two GPT partition flags: one may be set on a partition to
11165 cause it to be mounted read-only, and the other may be set
11166 on a partition to ignore it during automatic discovery.
11168 * Two new GPT type UUIDs have been added for automatic root
11169 partition discovery, for 32-bit and 64-bit ARM. This is not
11170 particularly useful for discovering the root directory on
11171 these architectures during bare-metal boots (since UEFI is
11172 not common there), but still very useful to allow booting of
11173 ARM disk images in nspawn with the -i option.
11175 * MAC addresses of interfaces created with nspawn's
11176 --network-interface= switch will now be generated from the
11177 machine name, and thus be stable between multiple invocations
11180 * logind will now automatically remove all IPC objects owned
11181 by a user if she or he fully logs out. This makes sure that
11182 users who are logged out cannot continue to consume IPC
11183 resources. This covers SysV memory, semaphores and message
11184 queues as well as POSIX shared memory and message
11185 queues. Traditionally, SysV and POSIX IPC had no lifecycle
11186 limits. With this functionality, that is corrected. This may
11187 be turned off by using the RemoveIPC= switch of logind.conf.
11189 * The systemd-machine-id-setup and tmpfiles tools gained a
11190 --root= switch to operate on a specific root directory,
11193 * journald can now forward logged messages to the TTYs of all
11194 logged in users ("wall"). This is the default for all
11195 emergency messages now.
11197 * A new tool systemd-journal-remote has been added to stream
11198 journal log messages across the network.
11200 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
11201 controller trees are mounted into it. Note that the
11202 directories mounted beneath it are not read-only. This is a
11203 security measure and is particularly useful because glibc
11204 actually includes a search logic to pick any tmpfs it can
11205 find to implement shm_open() if /dev/shm is not available
11206 (which it might very well be in namespaced setups).
11208 * machinectl gained a new "poweroff" command to cleanly power
11209 down a local OS container.
11211 * The PrivateDevices= unit file setting will now also drop the
11212 CAP_MKNOD capability from the capability bound set, and
11213 imply DevicePolicy=closed.
11215 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
11216 comprehensively on all long-running systemd services where
11217 this is appropriate.
11219 * systemd-udevd will now run in a disassociated mount
11220 namespace. To mount directories from udev rules, make sure to
11221 pull in mount units via SYSTEMD_WANTS properties.
11223 * The kdbus support gained support for uploading policy into
11224 the kernel. sd-bus gained support for creating "monitoring"
11225 connections that can eavesdrop into all bus communication
11226 for debugging purposes.
11228 * Timestamps may now be specified in seconds since the UNIX
11229 epoch Jan 1st, 1970 by specifying "@" followed by the value
11232 * Native tcpwrap support in systemd has been removed. tcpwrap
11233 is old code, not really maintained anymore and has serious
11234 shortcomings, and better options such as firewalls
11235 exist. For setups that require tcpwrap usage, please
11236 consider invoking your socket-activated service via tcpd,
11237 like on traditional inetd.
11239 * A new system.conf configuration option
11240 DefaultTimerAccuracySec= has been added that controls the
11241 default AccuracySec= setting of .timer units.
11243 * Timer units gained a new WakeSystem= switch. If enabled,
11244 timers configured this way will cause the system to resume
11245 from system suspend (if the system supports that, which most
11248 * Timer units gained a new Persistent= switch. If enabled,
11249 timers configured this way will save to disk when they have
11250 been last triggered. This information is then used on next
11251 reboot to possible execute overdue timer events, that
11252 could not take place because the system was powered off.
11253 This enables simple anacron-like behaviour for timer units.
11255 * systemctl's "list-timers" will now also list the time a
11256 timer unit was last triggered in addition to the next time
11257 it will be triggered.
11259 * systemd-networkd will now assign predictable IPv4LL
11260 addresses to its local interfaces.
11262 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
11263 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
11264 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
11265 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
11266 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
11267 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
11268 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
11269 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
11272 — Berlin, 2014-03-25
11276 * A new unit file setting RestrictAddressFamilies= has been
11277 added to restrict which socket address families unit
11278 processes gain access to. This takes address family names
11279 like "AF_INET" or "AF_UNIX", and is useful to minimize the
11280 attack surface of services via exotic protocol stacks. This
11281 is built on seccomp system call filters.
11283 * Two new unit file settings RuntimeDirectory= and
11284 RuntimeDirectoryMode= have been added that may be used to
11285 manage a per-daemon runtime directories below /run. This is
11286 an alternative for setting up directory permissions with
11287 tmpfiles snippets, and has the advantage that the runtime
11288 directory's lifetime is bound to the daemon runtime and that
11289 the daemon starts up with an empty directory each time. This
11290 is particularly useful when writing services that drop
11291 privileges using the User= or Group= setting.
11293 * The DeviceAllow= unit setting now supports globbing for
11294 matching against device group names.
11296 * The systemd configuration file system.conf gained new
11297 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
11298 DefaultMemoryAccounting= to globally turn on/off accounting
11299 for specific resources (cgroups) for all units. These
11300 settings may still be overridden individually in each unit
11303 * systemd-gpt-auto-generator is now able to discover /srv and
11304 root partitions in addition to /home and swap partitions. It
11305 also supports LUKS-encrypted partitions now. With this in
11306 place, automatic discovery of partitions to mount following
11307 the Discoverable Partitions Specification
11308 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
11309 is now a lot more complete. This allows booting without
11310 /etc/fstab and without root= on the kernel command line on
11311 systems prepared appropriately.
11313 * systemd-nspawn gained a new --image= switch which allows
11314 booting up disk images and Linux installations on any block
11315 device that follow the Discoverable Partitions Specification
11316 (see above). This means that installations made with
11317 appropriately updated installers may now be started and
11318 deployed using container managers, completely
11319 unmodified. (We hope that libvirt-lxc will add support for
11320 this feature soon, too.)
11322 * systemd-nspawn gained a new --network-macvlan= setting to
11323 set up a private macvlan interface for the
11324 container. Similarly, systemd-networkd gained a new
11325 Kind=macvlan setting in .netdev files.
11327 * systemd-networkd now supports configuring local addresses
11330 * A new tool systemd-network-wait-online has been added to
11331 synchronously wait for network connectivity using
11334 * The sd-bus.h bus API gained a new sd_bus_track object for
11335 tracking the lifecycle of bus peers. Note that sd-bus.h is
11336 still not a public API though (unless you specify
11337 --enable-kdbus on the configure command line, which however
11338 voids your warranty and you get no API stability guarantee).
11340 * The $XDG_RUNTIME_DIR runtime directories for each user are
11341 now individual tmpfs instances, which has the benefit of
11342 introducing separate pools for each user, with individual
11343 size limits, and thus making sure that unprivileged clients
11344 can no longer negatively impact the system or other users by
11345 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
11346 RuntimeDirectorySize= has been introduced that allows
11347 controlling the default size limit for all users. It
11348 defaults to 10% of the available physical memory. This is no
11349 replacement for quotas on tmpfs though (which the kernel
11350 still does not support), as /dev/shm and /tmp are still
11351 shared resources used by both the system and unprivileged
11354 * logind will now automatically turn off automatic suspending
11355 on laptop lid close when more than one display is
11356 connected. This was previously expected to be implemented
11357 individually in desktop environments (such as GNOME),
11358 however has been added to logind now, in order to fix a
11359 boot-time race where a desktop environment might not have
11360 been started yet and thus not been able to take an inhibitor
11361 lock at the time where logind already suspends the system
11362 due to a closed lid.
11364 * logind will now wait at least 30s after each system
11365 suspend/resume cycle, and 3min after system boot before
11366 suspending the system due to a closed laptop lid. This
11367 should give USB docking stations and similar enough time to
11368 be probed and configured after system resume and boot in
11369 order to then act as suspend blocker.
11371 * systemd-run gained a new --property= setting which allows
11372 initialization of resource control properties (and others)
11373 for the created scope or service unit. Example: "systemd-run
11374 --property=BlockIOWeight=10 updatedb" may be used to run
11375 updatedb at a low block IO scheduling weight.
11377 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
11378 now also work in --scope mode.
11380 * When systemd is compiled with kdbus support, basic support
11381 for enforced policies is now in place. (Note that enabling
11382 kdbus still voids your warranty and no API compatibility
11383 promises are made.)
11385 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
11386 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
11387 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
11388 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
11389 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
11390 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
11391 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
11392 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
11393 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
11394 Zbigniew Jędrzejewski-Szmek
11396 — Berlin, 2014-03-12
11400 * systemd will now relabel /dev after loading the SMACK policy
11401 according to SMACK rules.
11403 * A new unit file option AppArmorProfile= has been added to
11404 set the AppArmor profile for the processes of a unit.
11406 * A new condition check ConditionArchitecture= has been added
11407 to conditionalize units based on the system architecture, as
11408 reported by uname()'s "machine" field.
11410 * systemd-networkd now supports matching on the system
11411 virtualization, architecture, kernel command line, hostname
11414 * logind is now a lot more aggressive when suspending the
11415 machine due to a closed laptop lid. Instead of acting only
11416 on the lid close action, it will continuously watch the lid
11417 status and act on it. This is useful for laptops where the
11418 power button is on the outside of the chassis so that it can
11419 be reached without opening the lid (such as the Lenovo
11420 Yoga). On those machines, logind will now immediately
11421 re-suspend the machine if the power button has been
11422 accidentally pressed while the laptop was suspended and in a
11423 backpack or similar.
11425 * logind will now watch SW_DOCK switches and inhibit reaction
11426 to the lid switch if it is pressed. This means that logind
11427 will not suspend the machine anymore if the lid is closed
11428 and the system is docked, if the laptop supports SW_DOCK
11429 notifications via the input layer. Note that ACPI docking
11430 stations do not generate this currently. Also note that this
11431 logic is usually not fully sufficient and Desktop
11432 Environments should take a lid switch inhibitor lock when an
11433 external display is connected, as systemd will not watch
11436 * nspawn will now make use of the devices cgroup controller by
11437 default, and only permit creation of and access to the usual
11438 API device nodes like /dev/null or /dev/random, as well as
11439 access to (but not creation of) the pty devices.
11441 * We will now ship a default .network file for
11442 systemd-networkd that automatically configures DHCP for
11443 network interfaces created by nspawn's --network-veth or
11444 --network-bridge= switches.
11446 * systemd will now understand the usual M, K, G, T suffixes
11447 according to SI conventions (i.e. to the base 1000) when
11448 referring to throughput and hardware metrics. It will stay
11449 with IEC conventions (i.e. to the base 1024) for software
11450 metrics, according to what is customary according to
11451 Wikipedia. We explicitly document which base applies for
11452 each configuration option.
11454 * The DeviceAllow= setting in unit files now supports a syntax to
11455 allow-list an entire group of devices node majors at once, based on
11456 the /proc/devices listing. For example, with the string "char-pts",
11457 it is now possible to allow-list all current and future pseudo-TTYs
11460 * sd-event learned a new "post" event source. Event sources of
11461 this type are triggered by the dispatching of any event
11462 source of a type that is not "post". This is useful for
11463 implementing clean-up and check event sources that are
11464 triggered by other work being done in the program.
11466 * systemd-networkd is no longer statically enabled, but uses
11467 the usual [Install] sections so that it can be
11468 enabled/disabled using systemctl. It still is enabled by
11471 * When creating a veth interface pair with systemd-nspawn, the
11472 host side will now be prefixed with "vb-" if
11473 --network-bridge= is used, and with "ve-" if --network-veth
11474 is used. This way, it is easy to distinguish these cases on
11475 the host, for example to apply different configuration to
11476 them with systemd-networkd.
11478 * The compatibility libraries for libsystemd-journal.so,
11479 libsystem-id128.so, libsystemd-login.so and
11480 libsystemd-daemon.so do not make use of IFUNC
11481 anymore. Instead, we now build libsystemd.so multiple times
11482 under these alternative names. This means that the footprint
11483 is drastically increased, but given that these are
11484 transitional compatibility libraries, this should not matter
11485 much. This change has been made necessary to support the ARM
11486 platform for these compatibility libraries, as the ARM
11487 toolchain is not really at the same level as the toolchain
11488 for other architectures like x86 and does not support
11489 IFUNC. Please make sure to use --enable-compat-libs only
11490 during a transitional period!
11492 * The .include syntax has been deprecated and is not documented
11493 anymore. Drop-in files in .d directories should be used instead.
11495 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
11496 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
11497 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
11498 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
11499 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
11500 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
11501 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
11502 Zbigniew Jędrzejewski-Szmek
11504 — Berlin, 2014-02-24
11508 * A new component "systemd-networkd" has been added that can
11509 be used to configure local network interfaces statically or
11510 via DHCP. It is capable of bringing up bridges, VLANs, and
11511 bonding. Currently, no hook-ups for interactive network
11512 configuration are provided. Use this for your initrd,
11513 container, embedded, or server setup if you need a simple,
11514 yet powerful, network configuration solution. This
11515 configuration subsystem is quite nifty, as it allows wildcard
11516 hotplug matching in interfaces. For example, with a single
11517 configuration snippet, you can configure that all Ethernet
11518 interfaces showing up are automatically added to a bridge,
11519 or similar. It supports link-sensing and more.
11521 * A new tool "systemd-socket-proxyd" has been added which can
11522 act as a bidirectional proxy for TCP sockets. This is
11523 useful for adding socket activation support to services that
11524 do not actually support socket activation, including virtual
11525 machines and the like.
11527 * Add a new tool to save/restore rfkill state on
11530 * Save/restore state of keyboard backlights in addition to
11531 display backlights on shutdown/boot.
11533 * udev learned a new SECLABEL{} construct to label device
11534 nodes with a specific security label when they appear. For
11535 now, only SECLABEL{selinux} is supported, but the syntax is
11536 prepared for additional security frameworks.
11538 * udev gained a new scheme to configure link-level attributes
11539 from files in /etc/systemd/network/*.link. These files can
11540 match against MAC address, device path, driver name and type,
11541 and will apply attributes like the naming policy, link speed,
11542 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
11543 address assignment policy (randomized, …).
11545 * The configuration of network interface naming rules for
11546 "permanent interface names" has changed: a new NamePolicy=
11547 setting in the [Link] section of .link files determines the
11548 priority of possible naming schemes (onboard, slot, MAC,
11549 path). The default value of this setting is determined by
11550 /usr/lib/net/links/99-default.link. Old
11551 80-net-name-slot.rules udev configuration file has been
11552 removed, so local configuration overriding this file should
11553 be adapted to override 99-default.link instead.
11555 * When the User= switch is used in a unit file, also
11556 initialize $SHELL= based on the user database entry.
11558 * systemd no longer depends on libdbus. All communication is
11559 now done with sd-bus, systemd's low-level bus library
11562 * kdbus support has been added to PID 1 itself. When kdbus is
11563 enabled, this causes PID 1 to set up the system bus and
11564 enable support for a new ".busname" unit type that
11565 encapsulates bus name activation on kdbus. It works a little
11566 bit like ".socket" units, except for bus names. A new
11567 generator has been added that converts classic dbus1 service
11568 activation files automatically into native systemd .busname
11569 and .service units.
11571 * sd-bus: add a light-weight vtable implementation that allows
11572 defining objects on the bus with a simple static const
11573 vtable array of its methods, signals and properties.
11575 * systemd will not generate or install static dbus
11576 introspection data anymore to /usr/share/dbus-1/interfaces,
11577 as the precise format of these files is unclear, and
11578 nothing makes use of it.
11580 * A proxy daemon is now provided to proxy clients connecting
11581 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
11582 compatibility with classic D-Bus.
11584 * A bus driver implementation has been added that supports the
11585 classic D-Bus bus driver calls on kdbus, also for
11586 compatibility purposes.
11588 * A new API "sd-event.h" has been added that implements a
11589 minimal event loop API built around epoll. It provides a
11590 couple of features that direct epoll usage is lacking:
11591 prioritization of events, scales to large numbers of timer
11592 events, per-event timer slack (accuracy), system-wide
11593 coalescing of timer events, exit handlers, watchdog
11594 supervision support using systemd's sd_notify() API, child
11597 * A new API "sd-rntl.h" has been added that provides an API
11598 around the route netlink interface of the kernel, similar in
11599 style to "sd-bus.h".
11601 * A new API "sd-dhcp-client.h" has been added that provides a
11602 small DHCPv4 client-side implementation. This is used by
11603 "systemd-networkd".
11605 * There is a new kernel command line option
11606 "systemd.restore_state=0|1". When set to "0", none of the
11607 systemd tools will restore saved runtime state to hardware
11608 devices. More specifically, the rfkill and backlight states
11611 * The FsckPassNo= compatibility option in mount/service units
11612 has been removed. The fstab generator will now add the
11613 necessary dependencies automatically, and does not require
11614 PID1's support for that anymore.
11616 * journalctl gained a new switch, --list-boots, that lists
11617 recent boots with their times and boot IDs.
11619 * The various tools like systemctl, loginctl, timedatectl,
11620 busctl, systemd-run, … have gained a new switch "-M" to
11621 connect to a specific, local OS container (as direct
11622 connection, without requiring SSH). This works on any
11623 container that is registered with machined, such as those
11624 created by libvirt-lxc or nspawn.
11626 * systemd-run and systemd-analyze also gained support for "-H"
11627 to connect to remote hosts via SSH. This is particularly
11628 useful for systemd-run because it enables queuing of jobs
11629 onto remote systems.
11631 * machinectl gained a new command "login" to open a getty
11632 login in any local container. This works with any container
11633 that is registered with machined (such as those created by
11634 libvirt-lxc or nspawn), and which runs systemd inside.
11636 * machinectl gained a new "reboot" command that may be used to
11637 trigger a reboot on a specific container that is registered
11638 with machined. This works on any container that runs an init
11639 system of some kind.
11641 * systemctl gained a new "list-timers" command to print a nice
11642 listing of installed timer units with the times they elapse
11645 * Alternative reboot() parameters may now be specified on the
11646 "systemctl reboot" command line and are passed to the
11647 reboot() system call.
11649 * systemctl gained a new --job-mode= switch to configure the
11650 mode to queue a job with. This is a more generic version of
11651 --fail, --irreversible, and --ignore-dependencies, which are
11652 still available but not advertised anymore.
11654 * /etc/systemd/system.conf gained new settings to configure
11655 various default timeouts of units, as well as the default
11656 start limit interval and burst. These may still be overridden
11659 * PID1 will now export on the bus profile data of the security
11660 policy upload process (such as the SELinux policy upload to
11663 * journald: when forwarding logs to the console, include
11664 timestamps (following the setting in
11665 /sys/module/printk/parameters/time).
11667 * OnCalendar= in timer units now understands the special
11668 strings "yearly" and "annually". (Both are equivalent)
11670 * The accuracy of timer units is now configurable with the new
11671 AccuracySec= setting. It defaults to 1min.
11673 * A new dependency type JoinsNamespaceOf= has been added that
11674 allows running two services within the same /tmp and network
11675 namespace, if PrivateNetwork= or PrivateTmp= are used.
11677 * A new command "cat" has been added to systemctl. It outputs
11678 the original unit file of a unit, and concatenates the
11679 contents of additional "drop-in" unit file snippets, so that
11680 the full configuration is shown.
11682 * systemctl now supports globbing on the various "list-xyz"
11683 commands, like "list-units" or "list-sockets", as well as on
11684 those commands which take multiple unit names.
11686 * journalctl's --unit= switch gained support for globbing.
11688 * All systemd daemons now make use of the watchdog logic so
11689 that systemd automatically notices when they hang.
11691 * If the $container_ttys environment variable is set,
11692 getty-generator will automatically spawn a getty for each
11693 listed tty. This is useful for container managers to request
11694 login gettys to be spawned on as many ttys as needed.
11696 * %h, %s, %U specifier support is not available anymore when
11697 used in unit files for PID 1. This is because NSS calls are
11698 not safe from PID 1. They stay available for --user
11699 instances of systemd, and as special case for the root user.
11701 * loginctl gained a new "--no-legend" switch to turn off output
11702 of the legend text.
11704 * The "sd-login.h" API gained three new calls:
11705 sd_session_is_remote(), sd_session_get_remote_user(),
11706 sd_session_get_remote_host() to query information about
11709 * The udev hardware database now also carries vendor/product
11710 information of SDIO devices.
11712 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
11713 determine whether watchdog notifications are requested by
11714 the system manager.
11716 * Socket-activated per-connection services now include a
11717 short description of the connection parameters in the
11720 * tmpfiles gained a new "--boot" option. When this is not used,
11721 only lines where the command character is not suffixed with
11722 "!" are executed. When this option is specified, those
11723 options are executed too. This partitions tmpfiles
11724 directives into those that can be safely executed at any
11725 time, and those which should be run only at boot (for
11726 example, a line that creates /run/nologin).
11728 * A new API "sd-resolve.h" has been added which provides a simple
11729 asynchronous wrapper around glibc NSS hostname resolution
11730 calls, such as getaddrinfo(). In contrast to glibc's
11731 getaddrinfo_a(), it does not use signals. In contrast to most
11732 other asynchronous name resolution libraries, this one does
11733 not reimplement DNS, but reuses NSS, so that alternate
11734 hostname resolution systems continue to work, such as mDNS,
11735 LDAP, etc. This API is based on libasyncns, but it has been
11736 cleaned up for inclusion in systemd.
11738 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
11739 "sd-daemon.h" are no longer found in individual libraries
11740 libsystemd-journal.so, libsystemd-login.so,
11741 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
11742 merged them into a single library, libsystemd.so, which
11743 provides all symbols. The reason for this is cyclic
11744 dependencies, as these libraries tend to use each other's
11745 symbols. So far, we have managed to workaround that by linking
11746 a copy of a good part of our code into each of these
11747 libraries again and again, which, however, makes certain
11748 things hard to do, like sharing static variables. Also, it
11749 substantially increases footprint. With this change, there
11750 is only one library for the basic APIs systemd
11751 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
11752 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
11753 library as well, however are subject to the --enable-kdbus
11754 switch (see below). Note that "sd-dhcp-client.h" is not part
11755 of this library (this is because it only consumes, never
11756 provides, services of/to other APIs). To make the transition
11757 easy from the separate libraries to the unified one, we
11758 provide the --enable-compat-libs compile-time switch which
11759 will generate stub libraries that are compatible with the
11760 old ones but redirect all calls to the new one.
11762 * All of the kdbus logic and the new APIs "sd-bus.h",
11763 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
11764 and "sd-utf8.h" are compile-time optional via the
11765 "--enable-kdbus" switch, and they are not compiled in by
11766 default. To make use of kdbus, you have to explicitly enable
11767 the switch. Note however, that neither the kernel nor the
11768 userspace API for all of this is considered stable yet. We
11769 want to maintain the freedom to still change the APIs for
11770 now. By specifying this build-time switch, you acknowledge
11771 that you are aware of the instability of the current
11774 * Also, note that while kdbus is pretty much complete,
11775 it lacks one thing: proper policy support. This means you
11776 can build a fully working system with all features; however,
11777 it will be highly insecure. Policy support will be added in
11778 one of the next releases, at the same time that we will
11779 declare the APIs stable.
11781 * When the kernel command line argument "kdbus" is specified,
11782 systemd will automatically load the kdbus.ko kernel module. At
11783 this stage of development, it is only useful for testing kdbus
11784 and should not be used in production. Note: if "--enable-kdbus"
11785 is specified, and the kdbus.ko kernel module is available, and
11786 "kdbus" is added to the kernel command line, the entire system
11787 runs with kdbus instead of dbus-daemon, with the above mentioned
11788 problem of missing the system policy enforcement. Also a future
11789 version of kdbus.ko or a newer systemd will not be compatible with
11790 each other, and will unlikely be able to boot the machine if only
11791 one of them is updated.
11793 * systemctl gained a new "import-environment" command which
11794 uploads the caller's environment (or parts thereof) into the
11795 service manager so that it is inherited by services started
11796 by the manager. This is useful to upload variables like
11797 $DISPLAY into the user service manager.
11799 * A new PrivateDevices= switch has been added to service units
11800 which allows running a service with a namespaced /dev
11801 directory that does not contain any device nodes for
11802 physical devices. More specifically, it only includes devices
11803 such as /dev/null, /dev/urandom, and /dev/zero which are API
11806 * logind has been extended to support behaviour like VT
11807 switching on seats that do not support a VT. This makes
11808 multi-session available on seats that are not the first seat
11809 (seat0), and on systems where kernel support for VTs has
11810 been disabled at compile-time.
11812 * If a process holds a delay lock for system sleep or shutdown
11813 and fails to release it in time, we will now log its
11814 identity. This makes it easier to identify processes that
11815 cause slow suspends or power-offs.
11817 * When parsing /etc/crypttab, support for a new key-slot=
11818 option as supported by Debian is added. It allows indicating
11819 which LUKS slot to use on disk, speeding up key loading.
11821 * The sd_journal_sendv() API call has been checked and
11822 officially declared to be async-signal-safe so that it may
11823 be invoked from signal handlers for logging purposes.
11825 * Boot-time status output is now enabled automatically after a
11826 short timeout if boot does not progress, in order to give
11827 the user an indication what she or he is waiting for.
11829 * The boot-time output has been improved to show how much time
11830 remains until jobs expire.
11832 * The KillMode= switch in service units gained a new possible
11833 value "mixed". If set, and the unit is shut down, then the
11834 initial SIGTERM signal is sent only to the main daemon
11835 process, while the following SIGKILL signal is sent to
11836 all remaining processes of the service.
11838 * When a scope unit is registered, a new property "Controller"
11839 may be set. If set to a valid bus name, systemd will send a
11840 RequestStop() signal to this name when it would like to shut
11841 down the scope. This may be used to hook manager logic into
11842 the shutdown logic of scope units. Also, scope units may now
11843 be put in a special "abandoned" state, in which case the
11844 manager process which created them takes no further
11845 responsibilities for it.
11847 * When reading unit files, systemd will now verify
11848 the access mode of these files, and warn about certain
11849 suspicious combinations. This has been added to make it
11850 easier to track down packaging bugs where unit files are
11851 marked executable or world-writable.
11853 * systemd-nspawn gained a new "--setenv=" switch to set
11854 container-wide environment variables. The similar option in
11855 systemd-activate was renamed from "--environment=" to
11856 "--setenv=" for consistency.
11858 * systemd-nspawn has been updated to create a new kdbus domain
11859 for each container that is invoked, thus allowing each
11860 container to have its own set of system and user buses,
11861 independent of the host.
11863 * systemd-nspawn gained a new --drop-capability= switch to run
11864 the container with less capabilities than the default. Both
11865 --drop-capability= and --capability= now take the special
11866 string "all" for dropping or keeping all capabilities.
11868 * systemd-nspawn gained new switches for executing containers
11869 with specific SELinux labels set.
11871 * systemd-nspawn gained a new --quiet switch to not generate
11872 any additional output but the container's own console
11875 * systemd-nspawn gained a new --share-system switch to run a
11876 container without PID namespacing enabled.
11878 * systemd-nspawn gained a new --register= switch to control
11879 whether the container is registered with systemd-machined or
11880 not. This is useful for containers that do not run full
11881 OS images, but only specific apps.
11883 * systemd-nspawn gained a new --keep-unit which may be used
11884 when invoked as the only program from a service unit, and
11885 results in registration of the unit service itself in
11886 systemd-machined, instead of a newly opened scope unit.
11888 * systemd-nspawn gained a new --network-interface= switch for
11889 moving arbitrary interfaces to the container. The new
11890 --network-veth switch creates a virtual Ethernet connection
11891 between host and container. The new --network-bridge=
11892 switch then allows assigning the host side of this virtual
11893 Ethernet connection to a bridge device.
11895 * systemd-nspawn gained a new --personality= switch for
11896 setting the kernel personality for the container. This is
11897 useful when running a 32-bit container on a 64-bit host. A
11898 similar option Personality= is now also available for service
11901 * logind will now also track a "Desktop" identifier for each
11902 session which encodes the desktop environment of it. This is
11903 useful for desktop environments that want to identify
11904 multiple running sessions of itself easily.
11906 * A new SELinuxContext= setting for service units has been
11907 added that allows setting a specific SELinux execution
11908 context for a service.
11910 * Most systemd client tools will now honour $SYSTEMD_LESS for
11911 settings of the "less" pager. By default, these tools will
11912 override $LESS to allow certain operations to work, such as
11913 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
11914 influence this logic.
11916 * systemd's "seccomp" hook-up has been changed to make use of
11917 the libseccomp library instead of using its own
11918 implementation. This has benefits for portability among
11921 * For usage together with SystemCallFilter=, a new
11922 SystemCallErrorNumber= setting has been introduced that
11923 allows configuration of a system error number to be returned
11924 on filtered system calls, instead of immediately killing the
11925 process. Also, SystemCallArchitectures= has been added to
11926 limit access to system calls of a particular architecture
11927 (in order to turn off support for unused secondary
11928 architectures). There is also a global
11929 SystemCallArchitectures= setting in system.conf now to turn
11930 off support for non-native system calls system-wide.
11932 * systemd requires a kernel with a working name_to_handle_at(),
11933 please see the kernel config requirements in the README file.
11935 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
11936 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
11937 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
11938 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
11939 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
11940 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
11941 Elia Pinto, Florian Weimer, George McCollister, Goffredo
11942 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
11943 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
11944 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
11945 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
11946 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
11947 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
11948 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
11949 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
11950 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
11951 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
11952 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
11953 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
11954 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
11955 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
11956 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
11957 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
11958 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
11960 — Berlin, 2014-02-20
11964 * logind has gained support for facilitating privileged input
11965 and drm device access for unprivileged clients. This work is
11966 useful to allow Wayland display servers (and similar
11967 programs, such as kmscon) to run under the user's ID and
11968 access input and drm devices which are normally
11969 protected. When this is used (and the kernel is new enough)
11970 logind will "mute" IO on the file descriptors passed to
11971 Wayland as long as it is in the background and "unmute" it
11972 if it returns into the foreground. This allows secure
11973 session switching without allowing background sessions to
11974 eavesdrop on input and display data. This also introduces
11975 session switching support if VT support is turned off in the
11976 kernel, and on seats that are not seat0.
11978 * A new kernel command line option luks.options= is understood
11979 now which allows specifying LUKS options for usage for LUKS
11980 encrypted partitions specified with luks.uuid=.
11982 * tmpfiles.d(5) snippets may now use specifier expansion in
11983 path names. More specifically %m, %b, %H, %v, are now
11984 replaced by the local machine id, boot id, hostname, and
11985 kernel version number.
11987 * A new tmpfiles.d(5) command "m" has been introduced which
11988 may be used to change the owner/group/access mode of a file
11989 or directory if it exists, but do nothing if it does not.
11991 * This release removes high-level support for the
11992 MemorySoftLimit= cgroup setting. The underlying kernel
11993 cgroup attribute memory.soft_limit= is currently badly
11994 designed and likely to be removed from the kernel API in its
11995 current form, hence we should not expose it for now.
11997 * The memory.use_hierarchy cgroup attribute is now enabled for
11998 all cgroups systemd creates in the memory cgroup
11999 hierarchy. This option is likely to be come the built-in
12000 default in the kernel anyway, and the non-hierarchical mode
12001 never made much sense in the intrinsically hierarchical
12004 * A new field _SYSTEMD_SLICE= is logged along with all journal
12005 messages containing the slice a message was generated
12006 from. This is useful to allow easy per-customer filtering of
12007 logs among other things.
12009 * systemd-journald will no longer adjust the group of journal
12010 files it creates to the "systemd-journal" group. Instead we
12011 rely on the journal directory to be owned by the
12012 "systemd-journal" group, and its setgid bit set, so that the
12013 kernel file system layer will automatically enforce that
12014 journal files inherit this group assignment. The reason for
12015 this change is that we cannot allow NSS look-ups from
12016 journald which would be necessary to resolve
12017 "systemd-journal" to a numeric GID, because this might
12018 create deadlocks if NSS involves synchronous queries to
12019 other daemons (such as nscd, or sssd) which in turn are
12020 logging clients of journald and might block on it, which
12021 would then dead lock. A tmpfiles.d(5) snippet included in
12022 systemd will make sure the setgid bit and group are
12023 properly set on the journal directory if it exists on every
12024 boot. However, we recommend adjusting it manually after
12025 upgrades too (or from RPM scriptlets), so that the change is
12026 not delayed until next reboot.
12028 * Backlight and random seed files in /var/lib/ have moved into
12029 the /var/lib/systemd/ directory, in order to centralize all
12030 systemd generated files in one directory.
12032 * Boot time performance measurements (as displayed by
12033 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
12034 performance information if that's available to determine how
12035 much time BIOS and boot loader initialization required. With
12036 a sufficiently new BIOS you hence no longer need to boot
12037 with Gummiboot to get access to such information.
12039 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
12040 Cristian Rodríguez, Dave Reisner, David Herrmann, David
12041 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
12042 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
12043 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
12044 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
12045 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
12047 — Berlin, 2013-10-02
12051 * The Restart= option for services now understands a new
12052 on-watchdog setting, which will restart the service
12053 automatically if the service stops sending out watchdog keep
12054 alive messages (as configured with WatchdogSec=).
12056 * The getty generator (which is responsible for bringing up a
12057 getty on configured serial consoles) will no longer only
12058 start a getty on the primary kernel console but on all
12059 others, too. This makes the order in which console= is
12060 specified on the kernel command line less important.
12062 * libsystemd-logind gained a new sd_session_get_vt() call to
12063 retrieve the VT number of a session.
12065 * If the option "tries=0" is set for an entry of /etc/crypttab
12066 its passphrase is queried indefinitely instead of any
12067 maximum number of tries.
12069 * If a service with a configure PID file terminates its PID
12070 file will now be removed automatically if it still exists
12071 afterwards. This should put an end to stale PID files.
12073 * systemd-run will now also take relative binary path names
12074 for execution and no longer insists on absolute paths.
12076 * InaccessibleDirectories= and ReadOnlyDirectories= now take
12077 paths that are optionally prefixed with "-" to indicate that
12078 it should not be considered a failure if they do not exist.
12080 * journalctl -o (and similar commands) now understands a new
12081 output mode "short-precise", it is similar to "short" but
12082 shows timestamps with usec accuracy.
12084 * The option "discard" (as known from Debian) is now
12085 synonymous to "allow-discards" in /etc/crypttab. In fact,
12086 "discard" is preferred now (since it is easier to remember
12089 * Some licensing clean-ups were made, so that more code is now
12090 LGPL-2.1 licensed than before.
12092 * A minimal tool to save/restore the display backlight
12093 brightness across reboots has been added. It will store the
12094 backlight setting as late as possible at shutdown, and
12095 restore it as early as possible during reboot.
12097 * A logic to automatically discover and enable home and swap
12098 partitions on GPT disks has been added. With this in place
12099 /etc/fstab becomes optional for many setups as systemd can
12100 discover certain partitions located on the root disk
12101 automatically. Home partitions are recognized under their
12102 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
12103 partitions are recognized under their GPT type ID
12104 0657fd6da4ab43c484e50933c84b4f4f.
12106 * systemd will no longer pass any environment from the kernel
12107 or initrd to system services. If you want to set an
12108 environment for all services, do so via the kernel command
12109 line systemd.setenv= assignment.
12111 * The systemd-sysctl tool no longer natively reads the file
12112 /etc/sysctl.conf. If desired, the file should be symlinked
12113 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
12114 legacy support by a symlink rather than built-in code, it
12115 also makes the otherwise hidden order of application of the
12116 different files visible. (Note that this partly reverts to a
12117 pre-198 application order of sysctl knobs!)
12119 * The "systemctl set-log-level" and "systemctl dump" commands
12120 have been moved to systemd-analyze.
12122 * systemd-run learned the new --remain-after-exit switch,
12123 which causes the scope unit not to be cleaned up
12124 automatically after the process terminated.
12126 * tmpfiles learned a new --exclude-prefix= switch to exclude
12127 certain paths from operation.
12129 * journald will now automatically flush all messages to disk
12130 as soon as a message at the log level CRIT, ALERT or EMERG
12133 Contributions from: Andrew Cook, Brandon Philips, Christian
12134 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
12135 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
12136 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
12137 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
12138 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
12139 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
12140 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
12141 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
12142 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
12143 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
12144 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
12145 William Giokas, Zbigniew Jędrzejewski-Szmek
12147 — Berlin, 2013-09-13
12151 * The documentation has been updated to cover the various new
12152 concepts introduced with 205.
12154 * Unit files now understand the new %v specifier which
12155 resolves to the kernel version string as returned by "uname
12158 * systemctl now supports filtering the unit list output by
12159 load state, active state and sub state, using the new
12160 --state= parameter.
12162 * "systemctl status" will now show the results of the
12163 condition checks (like ConditionPathExists= and similar) of
12164 the last start attempts of the unit. They are also logged to
12167 * "journalctl -b" may now be used to look for boot output of a
12168 specific boot. Try "journalctl -b -1" for the previous boot,
12169 but the syntax is substantially more powerful.
12171 * "journalctl --show-cursor" has been added which prints the
12172 cursor string the last shown log line. This may then be used
12173 with the new "journalctl --after-cursor=" switch to continue
12174 browsing logs from that point on.
12176 * "journalctl --force" may now be used to force regeneration
12179 * Creation of "dead" device nodes has been moved from udev
12180 into kmod and tmpfiles. Previously, udev would read the kmod
12181 databases to pre-generate dead device nodes based on meta
12182 information contained in kernel modules, so that these would
12183 be auto-loaded on access rather then at boot. As this
12184 does not really have much to do with the exposing actual
12185 kernel devices to userspace this has always been slightly
12186 alien in the udev codebase. Following the new scheme kmod
12187 will now generate a runtime snippet for tmpfiles from the
12188 module meta information and it now is tmpfiles' job to the
12189 create the nodes. This also allows overriding access and
12190 other parameters for the nodes using the usual tmpfiles
12191 facilities. As side effect this allows us to remove the
12192 CAP_SYS_MKNOD capability bit from udevd entirely.
12194 * logind's device ACLs may now be applied to these "dead"
12195 devices nodes too, thus finally allowing managed access to
12196 devices such as /dev/snd/sequencer without loading the
12197 backing module right-away.
12199 * A new RPM macro has been added that may be used to apply
12200 tmpfiles configuration during package installation.
12202 * systemd-detect-virt and ConditionVirtualization= now can
12203 detect User-Mode-Linux machines (UML).
12205 * journald will now implicitly log the effective capabilities
12206 set of processes in the message metadata.
12208 * systemd-cryptsetup has gained support for TrueCrypt volumes.
12210 * The initrd interface has been simplified (more specifically,
12211 support for passing performance data via environment
12212 variables and fsck results via files in /run has been
12213 removed). These features were non-essential, and are
12214 nowadays available in a much nicer way by having systemd in
12215 the initrd serialize its state and have the hosts systemd
12216 deserialize it again.
12218 * The udev "keymap" data files and tools to apply keyboard
12219 specific mappings of scan to key codes, and force-release
12220 scan code lists have been entirely replaced by a udev
12221 "keyboard" builtin and a hwdb data file.
12223 * systemd will now honour the kernel's "quiet" command line
12224 argument also during late shutdown, resulting in a
12225 completely silent shutdown when used.
12227 * There's now an option to control the SO_REUSEPORT socket
12228 option in .socket units.
12230 * Instance units will now automatically get a per-template
12231 subslice of system.slice unless something else is explicitly
12232 configured. For example, instances of sshd@.service will now
12233 implicitly be placed in system-sshd.slice rather than
12234 system.slice as before.
12236 * Test coverage support may now be enabled at build time.
12238 Contributions from: Dave Reisner, Frederic Crozat, Harald
12239 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
12240 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
12241 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
12242 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
12243 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
12244 Giokas, Zbigniew Jędrzejewski-Szmek
12246 — Berlin, 2013-07-23
12250 * Two new unit types have been introduced:
12252 Scope units are very similar to service units, however, are
12253 created out of pre-existing processes — instead of PID 1
12254 forking off the processes. By using scope units it is
12255 possible for system services and applications to group their
12256 own child processes (worker processes) in a powerful way
12257 which then maybe used to organize them, or kill them
12258 together, or apply resource limits on them.
12260 Slice units may be used to partition system resources in an
12261 hierarchical fashion and then assign other units to them. By
12262 default there are now three slices: system.slice (for all
12263 system services), user.slice (for all user sessions),
12264 machine.slice (for VMs and containers).
12266 Slices and scopes have been introduced primarily in
12267 context of the work to move cgroup handling to a
12268 single-writer scheme, where only PID 1
12269 creates/removes/manages cgroups.
12271 * There's a new concept of "transient" units. In contrast to
12272 normal units these units are created via an API at runtime,
12273 not from configuration from disk. More specifically this
12274 means it is now possible to run arbitrary programs as
12275 independent services, with all execution parameters passed
12276 in via bus APIs rather than read from disk. Transient units
12277 make systemd substantially more dynamic then it ever was,
12278 and useful as a general batch manager.
12280 * logind has been updated to make use of scope and slice units
12281 for managing user sessions. As a user logs in he will get
12282 his own private slice unit, to which all sessions are added
12283 as scope units. We also added support for automatically
12284 adding an instance of user@.service for the user into the
12285 slice. Effectively logind will no longer create cgroup
12286 hierarchies on its own now, it will defer entirely to PID 1
12287 for this by means of scope, service and slice units. Since
12288 user sessions this way become entities managed by PID 1
12289 the output of "systemctl" is now a lot more comprehensive.
12291 * A new mini-daemon "systemd-machined" has been added which
12292 may be used by virtualization managers to register local
12293 VMs/containers. nspawn has been updated accordingly, and
12294 libvirt will be updated shortly. machined will collect a bit
12295 of meta information about the VMs/containers, and assign
12296 them their own scope unit (see above). The collected
12297 meta-data is then made available via the "machinectl" tool,
12298 and exposed in "ps" and similar tools. machined/machinectl
12299 is compile-time optional.
12301 * As discussed earlier, the low-level cgroup configuration
12302 options ControlGroup=, ControlGroupModify=,
12303 ControlGroupPersistent=, ControlGroupAttribute= have been
12304 removed. Please use high-level attribute settings instead as
12305 well as slice units.
12307 * A new bus call SetUnitProperties() has been added to alter
12308 various runtime parameters of a unit. This is primarily
12309 useful to alter cgroup parameters dynamically in a nice way,
12310 but will be extended later on to make more properties
12311 modifiable at runtime. systemctl gained a new set-properties
12312 command that wraps this call.
12314 * A new tool "systemd-run" has been added which can be used to
12315 run arbitrary command lines as transient services or scopes,
12316 while configuring a number of settings via the command
12317 line. This tool is currently very basic, however already
12318 very useful. We plan to extend this tool to even allow
12319 queuing of execution jobs with time triggers from the
12320 command line, similar in fashion to "at".
12322 * nspawn will now inform the user explicitly that kernels with
12323 audit enabled break containers, and suggest the user to turn
12326 * Support for detecting the IMA and AppArmor security
12327 frameworks with ConditionSecurity= has been added.
12329 * journalctl gained a new "-k" switch for showing only kernel
12330 messages, mimicking dmesg output; in addition to "--user"
12331 and "--system" switches for showing only user's own logs
12334 * systemd-delta can now show information about drop-in
12335 snippets extending unit files.
12337 * libsystemd-bus has been substantially updated but is still
12338 not available as public API.
12340 * systemd will now look for the "debug" argument on the kernel
12341 command line and enable debug logging, similar to what
12342 "systemd.log_level=debug" already did before.
12344 * "systemctl set-default", "systemctl get-default" has been
12345 added to configure the default.target symlink, which
12346 controls what to boot into by default.
12348 * "systemctl set-log-level" has been added as a convenient
12349 way to raise and lower systemd logging threshold.
12351 * "systemd-analyze plot" will now show the time the various
12352 generators needed for execution, as well as information
12353 about the unit file loading.
12355 * libsystemd-journal gained a new sd_journal_open_files() call
12356 for opening specific journal files. journactl also gained a
12357 new switch to expose this new functionality. Previously we
12358 only supported opening all files from a directory, or all
12359 files from the system, as opening individual files only is
12360 racy due to journal file rotation.
12362 * systemd gained the new DefaultEnvironment= setting in
12363 /etc/systemd/system.conf to set environment variables for
12366 * If a privileged process logs a journal message with the
12367 OBJECT_PID= field set, then journald will automatically
12368 augment this with additional OBJECT_UID=, OBJECT_GID=,
12369 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
12370 system services want to log events about specific client
12371 processes. journactl/systemctl has been updated to make use
12372 of this information if all log messages regarding a specific
12375 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
12376 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
12377 Reisner, David Coppa, David King, David Strauss, Eelco
12378 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
12379 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
12380 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
12381 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
12382 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
12383 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
12384 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
12385 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
12386 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
12387 Łukasz Stelmach, 장동준
12391 * The Python bindings gained some minimal support for the APIs
12392 exposed by libsystemd-logind.
12394 * ConditionSecurity= gained support for detecting SMACK. Since
12395 this condition already supports SELinux and AppArmor we only
12396 miss IMA for this. Patches welcome!
12398 Contributions from: Karol Lewandowski, Lennart Poettering,
12399 Zbigniew Jędrzejewski-Szmek
12403 * systemd-nspawn will now create /etc/resolv.conf if
12404 necessary, before bind-mounting the host's file onto it.
12406 * systemd-nspawn will now store meta information about a
12407 container on the container's cgroup as extended attribute
12408 fields, including the root directory.
12410 * The cgroup hierarchy has been reworked in many ways. All
12411 objects any of the components systemd creates in the cgroup
12412 tree are now suffixed. More specifically, user sessions are
12413 now placed in cgroups suffixed with ".session", users in
12414 cgroups suffixed with ".user", and nspawn containers in
12415 cgroups suffixed with ".nspawn". Furthermore, all cgroup
12416 names are now escaped in a simple scheme to avoid collision
12417 of userspace object names with kernel filenames. This work
12418 is preparation for making these objects relocatable in the
12419 cgroup tree, in order to allow easy resource partitioning of
12420 these objects without causing naming conflicts.
12422 * systemctl list-dependencies gained the new switches
12423 --plain, --reverse, --after and --before.
12425 * systemd-inhibit now shows the process name of processes that
12426 have taken an inhibitor lock.
12428 * nss-myhostname will now also resolve "localhost"
12429 implicitly. This makes /etc/hosts an optional file and
12430 nicely handles that on IPv6 ::1 maps to both "localhost" and
12431 the local hostname.
12433 * libsystemd-logind.so gained a new call
12434 sd_get_machine_names() to enumerate running containers and
12435 VMs (currently only supported by very new libvirt and
12436 nspawn). sd_login_monitor can now be used to watch
12437 VMs/containers coming and going.
12439 * .include is not allowed recursively anymore, and only in
12440 unit files. Usually it is better to use drop-in snippets in
12441 .d/*.conf anyway, as introduced with systemd 198.
12443 * systemd-analyze gained a new "critical-chain" command that
12444 determines the slowest chain of units run during system
12445 boot-up. It is very useful for tracking down where
12446 optimizing boot time is the most beneficial.
12448 * systemd will no longer allow manipulating service paths in
12449 the name=systemd:/system cgroup tree using ControlGroup= in
12450 units. (But is still fine with it in all other dirs.)
12452 * There's a new systemd-nspawn@.service service file that may
12453 be used to easily run nspawn containers as system
12454 services. With the container's root directory in
12455 /var/lib/container/foobar it is now sufficient to run
12456 "systemctl start systemd-nspawn@foobar.service" to boot it.
12458 * systemd-cgls gained a new parameter "--machine" to list only
12459 the processes within a certain container.
12461 * ConditionSecurity= now can check for "apparmor". We still
12462 are lacking checks for SMACK and IMA for this condition
12463 check though. Patches welcome!
12465 * A new configuration file /etc/systemd/sleep.conf has been
12466 added that may be used to configure which kernel operation
12467 systemd is supposed to execute when "suspend", "hibernate"
12468 or "hybrid-sleep" is requested. This makes the new kernel
12469 "freeze" state accessible to the user.
12471 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
12472 the passed argument if applicable.
12474 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
12475 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
12476 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
12477 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
12478 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
12479 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
12480 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
12485 * The output of 'systemctl list-jobs' got some polishing. The
12486 '--type=' argument may now be passed more than once. A new
12487 command 'systemctl list-sockets' has been added which shows
12488 a list of kernel sockets systemd is listening on with the
12489 socket units they belong to, plus the units these socket
12492 * The experimental libsystemd-bus library got substantial
12493 updates to work in conjunction with the (also experimental)
12494 kdbus kernel project. It works well enough to exchange
12495 messages with some sophistication. Note that kdbus is not
12496 ready yet, and the library is mostly an elaborate test case
12497 for now, and not installable.
12499 * systemd gained a new unit 'systemd-static-nodes.service'
12500 that generates static device nodes earlier during boot, and
12501 can run in conjunction with udev.
12503 * libsystemd-login gained a new call sd_pid_get_user_unit()
12504 to retrieve the user systemd unit a process is running
12505 in. This is useful for systems where systemd is used as
12508 * systemd-nspawn now places all containers in the new /machine
12509 top-level cgroup directory in the name=systemd
12510 hierarchy. libvirt will soon do the same, so that we get a
12511 uniform separation of /system, /user and /machine for system
12512 services, user processes and containers/virtual
12513 machines. This new cgroup hierarchy is also useful to stick
12514 stable names to specific container instances, which can be
12515 recognized later this way (this name may be controlled
12516 via systemd-nspawn's new -M switch). libsystemd-login also
12517 gained a new call sd_pid_get_machine_name() to retrieve the
12518 name of the container/VM a specific process belongs to.
12520 * bootchart can now store its data in the journal.
12522 * libsystemd-journal gained a new call
12523 sd_journal_add_conjunction() for AND expressions to the
12524 matching logic. This can be used to express more complex
12525 logical expressions.
12527 * journactl can now take multiple --unit= and --user-unit=
12530 * The cryptsetup logic now understands the "luks.key=" kernel
12531 command line switch for specifying a file to read the
12532 decryption key from. Also, if a configured key file is not
12533 found the tool will now automatically fall back to prompting
12536 * Python systemd.journal module was updated to wrap recently
12537 added functions from libsystemd-journal. The interface was
12538 changed to bring the low level interface in s.j._Reader
12539 closer to the C API, and the high level interface in
12540 s.j.Reader was updated to wrap and convert all data about
12543 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
12544 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
12545 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
12546 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
12547 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
12548 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
12552 * journalctl --update-catalog now understands a new --root=
12553 option to operate on catalogs found in a different root
12556 * During shutdown after systemd has terminated all running
12557 services a final killing loop kills all remaining left-over
12558 processes. We will now print the name of these processes
12559 when we send SIGKILL to them, since this usually indicates a
12562 * If /etc/crypttab refers to password files stored on
12563 configured mount points automatic dependencies will now be
12564 generated to ensure the specific mount is established first
12565 before the key file is attempted to be read.
12567 * 'systemctl status' will now show information about the
12568 network sockets a socket unit is listening on.
12570 * 'systemctl status' will also shown information about any
12571 drop-in configuration file for units. (Drop-In configuration
12572 files in this context are files such as
12573 /etc/systemd/system/foobar.service.d/*.conf)
12575 * systemd-cgtop now optionally shows summed up CPU times of
12576 cgroups. Press '%' while running cgtop to switch between
12577 percentage and absolute mode. This is useful to determine
12578 which cgroups use up the most CPU time over the entire
12579 runtime of the system. systemd-cgtop has also been updated
12580 to be 'pipeable' for processing with further shell tools.
12582 * 'hostnamectl set-hostname' will now allow setting of FQDN
12585 * The formatting and parsing of time span values has been
12586 changed. The parser now understands fractional expressions
12587 such as "5.5h". The formatter will now output fractional
12588 expressions for all time spans under 1min, i.e. "5.123456s"
12589 rather than "5s 123ms 456us". For time spans under 1s
12590 millisecond values are shown, for those under 1ms
12591 microsecond values are shown. This should greatly improve
12592 all time-related output of systemd.
12594 * libsystemd-login and libsystemd-journal gained new
12595 functions for querying the poll() events mask and poll()
12596 timeout value for integration into arbitrary event
12599 * localectl gained the ability to list available X11 keymaps
12600 (models, layouts, variants, options).
12602 * 'systemd-analyze dot' gained the ability to filter for
12603 specific units via shell-style globs, to create smaller,
12604 more useful graphs. I.e. it is now possible to create simple
12605 graphs of all the dependencies between only target units, or
12606 of all units that Avahi has dependencies with.
12608 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
12609 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
12610 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
12611 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
12612 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
12613 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
12614 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
12618 * The boot-time readahead implementation for rotating media
12619 will now read the read-ahead data in multiple passes which
12620 consist of all read requests made in equidistant time
12621 intervals. This means instead of strictly reading read-ahead
12622 data in its physical order on disk we now try to find a
12623 middle ground between physical and access time order.
12625 * /etc/os-release files gained a new BUILD_ID= field for usage
12626 on operating systems that provide continuous builds of OS
12629 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
12630 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
12631 William Douglas, Zbigniew Jędrzejewski-Szmek
12635 * systemd-python gained an API exposing libsystemd-daemon.
12637 * The SMACK setup logic gained support for uploading CIPSO
12640 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
12641 ReadOnlyDirectories= and InaccessibleDirectories= has
12642 changed. The private /tmp and /var/tmp directories are now
12643 shared by all processes of a service (which means
12644 ExecStartPre= may now leave data in /tmp that ExecStart= of
12645 the same service can still access). When a service is
12646 stopped its temporary directories are immediately deleted
12647 (normal clean-up with tmpfiles is still done in addition to
12650 * By default, systemd will now set a couple of sysctl
12651 variables in the kernel: the safe sysrq options are turned
12652 on, IP route verification is turned on, and source routing
12653 disabled. The recently added hardlink and softlink
12654 protection of the kernel is turned on. These settings should
12655 be reasonably safe, and good defaults for all new systems.
12657 * The predictable network naming logic may now be turned off
12658 with a new kernel command line switch: net.ifnames=0.
12660 * A new libsystemd-bus module has been added that implements a
12661 pretty complete D-Bus client library. For details see:
12663 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
12665 * journald will now explicitly flush the journal files to disk
12666 at the latest 5min after each write. The file will then also
12667 be marked offline until the next write. This should increase
12668 reliability in case of a crash. The synchronization delay
12669 can be configured via SyncIntervalSec= in journald.conf.
12671 * There's a new remote-fs-setup.target unit that can be used
12672 to pull in specific services when at least one remote file
12673 system is to be mounted.
12675 * There are new targets timers.target and paths.target as
12676 canonical targets to pull user timer and path units in
12677 from. This complements sockets.target with a similar
12678 purpose for socket units.
12680 * libudev gained a new call udev_device_set_attribute_value()
12681 to set sysfs attributes of a device.
12683 * The udev daemon now sets the default number of worker
12684 processes executed in parallel based on the number of available
12685 CPUs instead of the amount of available RAM. This is supposed
12686 to provide a more reliable default and limit a too aggressive
12687 parallelism for setups with 1000s of devices connected.
12689 Contributions from: Auke Kok, Colin Walters, Cristian
12690 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
12691 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
12692 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
12693 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
12694 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
12695 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
12696 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
12697 Zbigniew Jędrzejewski-Szmek
12701 * Configuration of unit files may now be extended via drop-in
12702 files without having to edit/override the unit files
12703 themselves. More specifically, if the administrator wants to
12704 change one value for a service file foobar.service he can
12705 now do so by dropping in a configuration snippet into
12706 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
12707 will load all these snippets and apply them on top of the
12708 main unit configuration file, possibly extending or
12709 overriding its settings. Using these drop-in snippets is
12710 generally nicer than the two earlier options for changing
12711 unit files locally: copying the files from
12712 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
12713 them there; or creating a new file in /etc/systemd/system/
12714 that incorporates the original one via ".include". Drop-in
12715 snippets into these .d/ directories can be placed in any
12716 directory systemd looks for units in, and the usual
12717 overriding semantics between /usr/lib, /etc and /run apply
12720 * Most unit file settings which take lists of items can now be
12721 reset by assigning the empty string to them. For example,
12722 normally, settings such as Environment=FOO=BAR append a new
12723 environment variable assignment to the environment block,
12724 each time they are used. By assigning Environment= the empty
12725 string the environment block can be reset to empty. This is
12726 particularly useful with the .d/*.conf drop-in snippets
12727 mentioned above, since this adds the ability to reset list
12728 settings from vendor unit files via these drop-ins.
12730 * systemctl gained a new "list-dependencies" command for
12731 listing the dependencies of a unit recursively.
12733 * Inhibitors are now honored and listed by "systemctl
12734 suspend", "systemctl poweroff" (and similar) too, not only
12735 GNOME. These commands will also list active sessions by
12738 * Resource limits (as exposed by the various control group
12739 controllers) can now be controlled dynamically at runtime
12740 for all units. More specifically, you can now use a command
12741 like "systemctl set-cgroup-attr foobar.service cpu.shares
12742 2000" to alter the CPU shares a specific service gets. These
12743 settings are stored persistently on disk, and thus allow the
12744 administrator to easily adjust the resource usage of
12745 services with a few simple commands. This dynamic resource
12746 management logic is also available to other programs via the
12747 bus. Almost any kernel cgroup attribute and controller is
12750 * systemd-vconsole-setup will now copy all font settings to
12751 all allocated VTs, where it previously applied them only to
12754 * libsystemd-login gained the new sd_session_get_tty() API
12757 * This release drops support for a few legacy or
12758 distribution-specific LSB facility names when parsing init
12759 scripts: $x-display-manager, $mail-transfer-agent,
12760 $mail-transport-agent, $mail-transfer-agent, $smtp,
12761 $null. Also, the mail-transfer-agent.target unit backing
12762 this has been removed. Distributions which want to retain
12763 compatibility with this should carry the burden for
12764 supporting this themselves and patch support for these back
12765 in, if they really need to. Also, the facilities $syslog and
12766 $local_fs are now ignored, since systemd does not support
12767 early-boot LSB init scripts anymore, and these facilities
12768 are implied anyway for normal services. syslog.target has
12771 * There are new bus calls on PID1's Manager object for
12772 cancelling jobs, and removing snapshot units. Previously,
12773 both calls were only available on the Job and Snapshot
12774 objects themselves.
12776 * systemd-journal-gatewayd gained SSL support.
12778 * The various "environment" files, such as /etc/locale.conf
12779 now support continuation lines with a backslash ("\") as
12780 last character in the line, similarly in style (but different)
12781 to how this is supported in shells.
12783 * For normal user processes the _SYSTEMD_USER_UNIT= field is
12784 now implicitly appended to every log entry logged. systemctl
12785 has been updated to filter by this field when operating on a
12786 user systemd instance.
12788 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
12789 CAP_AUDIT_CONTROL capabilities to the capabilities set for
12790 the container. This makes it easier to boot unmodified
12791 Fedora systems in a container, which however still requires
12792 audit=0 to be passed on the kernel command line. Auditing in
12793 kernel and userspace is unfortunately still too broken in
12794 context of containers, hence we recommend compiling it out
12795 of the kernel or using audit=0. Hopefully this will be fixed
12796 one day for good in the kernel.
12798 * nspawn gained the new --bind= and --bind-ro= parameters to
12799 bind mount specific directories from the host into the
12802 * nspawn will now mount its own devpts file system instance
12803 into the container, in order not to leak pty devices from
12804 the host into the container.
12806 * systemd will now read the firmware boot time performance
12807 information from the EFI variables, if the used boot loader
12808 supports this, and takes it into account for boot performance
12809 analysis via "systemd-analyze". This is currently supported
12810 only in conjunction with Gummiboot, but could be supported
12811 by other boot loaders too. For details see:
12813 https://systemd.io/BOOT_LOADER_INTERFACE
12815 * A new generator has been added that automatically mounts the
12816 EFI System Partition (ESP) to /boot, if that directory
12817 exists, is empty, and no other file system has been
12818 configured to be mounted there.
12820 * logind will now send out PrepareForSleep(false) out
12821 unconditionally, after coming back from suspend. This may be
12822 used by applications as asynchronous notification for
12823 system resume events.
12825 * "systemctl unlock-sessions" has been added, that allows
12826 unlocking the screens of all user sessions at once, similar
12827 to how "systemctl lock-sessions" already locked all users
12828 sessions. This is backed by a new D-Bus call UnlockSessions().
12830 * "loginctl seat-status" will now show the master device of a
12831 seat. (i.e. the device of a seat that needs to be around for
12832 the seat to be considered available, usually the graphics
12835 * tmpfiles gained a new "X" line type, that allows
12836 configuration of files and directories (with wildcards) that
12837 shall be excluded from automatic cleanup ("aging").
12839 * udev default rules set the device node permissions now only
12840 at "add" events, and do not change them any longer with a
12841 later "change" event.
12843 * The log messages for lid events and power/sleep keypresses
12844 now carry a message ID.
12846 * We now have a substantially larger unit test suite, but this
12847 continues to be work in progress.
12849 * udevadm hwdb gained a new --root= parameter to change the
12850 root directory to operate relative to.
12852 * logind will now issue a background sync() request to the kernel
12853 early at shutdown, so that dirty buffers are flushed to disk early
12854 instead of at the last moment, in order to optimize shutdown
12857 * A new bootctl tool has been added that is an interface for
12858 certain boot loader operations. This is currently a preview
12859 and is likely to be extended into a small mechanism daemon
12860 like timedated, localed, hostnamed, and can be used by
12861 graphical UIs to enumerate available boot options, and
12862 request boot into firmware operations.
12864 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
12865 the rest of the package. It also has been updated to work
12866 correctly in initrds.
12868 * polkit previously has been runtime optional, and is now also
12869 compile time optional via a configure switch.
12871 * systemd-analyze has been reimplemented in C. Also "systemctl
12872 dot" has moved into systemd-analyze.
12874 * "systemctl status" with no further parameters will now print
12875 the status of all active or failed units.
12877 * Operations such as "systemctl start" can now be executed
12878 with a new mode "--irreversible" which may be used to queue
12879 operations that cannot accidentally be reversed by a later
12880 job queuing. This is by default used to make shutdown
12881 requests more robust.
12883 * The Python API of systemd now gained a new module for
12884 reading journal files.
12886 * A new tool kernel-install has been added that can install
12887 kernel images according to the Boot Loader Specification:
12889 https://systemd.io/BOOT_LOADER_SPECIFICATION
12891 * Boot time console output has been improved to provide
12892 animated boot time output for hanging jobs.
12894 * A new tool systemd-activate has been added which can be used
12895 to test socket activation with, directly from the command
12896 line. This should make it much easier to test and debug
12897 socket activation in daemons.
12899 * journalctl gained a new "--reverse" (or -r) option to show
12900 journal output in reverse order (i.e. newest line first).
12902 * journalctl gained a new "--pager-end" (or -e) option to jump
12903 to immediately jump to the end of the journal in the
12904 pager. This is only supported in conjunction with "less".
12906 * journalctl gained a new "--user-unit=" option, that works
12907 similarly to "--unit=" but filters for user units rather than
12910 * A number of unit files to ease adoption of systemd in
12911 initrds has been added. This moves some minimal logic from
12912 the various initrd implementations into systemd proper.
12914 * The journal files are now owned by a new group
12915 "systemd-journal", which exists specifically to allow access
12916 to the journal, and nothing else. Previously, we used the
12917 "adm" group for that, which however possibly covers more
12918 than just journal/log file access. This new group is now
12919 already used by systemd-journal-gatewayd to ensure this
12920 daemon gets access to the journal files and as little else
12921 as possible. Note that "make install" will also set FS ACLs
12922 up for /var/log/journal to give "adm" and "wheel" read
12923 access to it, in addition to "systemd-journal" which owns
12924 the journal files. We recommend that packaging scripts also
12925 add read access to "adm" + "wheel" to /var/log/journal, and
12926 all existing/future journal files. To normal users and
12927 administrators little changes, however packagers need to
12928 ensure to create the "systemd-journal" system group at
12929 package installation time.
12931 * The systemd-journal-gatewayd now runs as unprivileged user
12932 systemd-journal-gateway:systemd-journal-gateway. Packaging
12933 scripts need to create these system user/group at
12936 * timedated now exposes a new boolean property CanNTP that
12937 indicates whether a local NTP service is available or not.
12939 * systemd-detect-virt will now also detect xen PVs
12941 * The pstore file system is now mounted by default, if it is
12944 * In addition to the SELinux and IMA policies we will now also
12945 load SMACK policies at early boot.
12947 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
12948 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
12949 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
12950 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
12951 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
12952 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
12953 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
12954 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
12955 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
12956 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
12957 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
12958 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
12959 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
12960 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
12964 * Timer units now support calendar time events in addition to
12965 monotonic time events. That means you can now trigger a unit
12966 based on a calendar time specification such as "Thu,Fri
12967 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
12968 or fifth day of any month of the year 2013, given that it is
12969 a Thursday or a Friday. This brings timer event support
12970 considerably closer to cron's capabilities. For details on
12971 the supported calendar time specification language see
12974 * udev now supports a number of different naming policies for
12975 network interfaces for predictable names, and a combination
12976 of these policies is now the default. Please see this wiki
12977 document for details:
12979 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
12981 * Auke Kok's bootchart implementation has been added to the
12982 systemd tree. It is an optional component that can graph the
12983 boot in quite some detail. It is one of the best bootchart
12984 implementations around and minimal in its code and
12987 * nss-myhostname has been integrated into the systemd source
12988 tree. nss-myhostname guarantees that the local hostname
12989 always stays resolvable via NSS. It has been a weak
12990 requirement of systemd-hostnamed since a long time, and
12991 since its code is actually trivial we decided to just
12992 include it in systemd's source tree. It can be turned off
12993 with a configure switch.
12995 * The read-ahead logic is now capable of properly detecting
12996 whether a btrfs file system is on SSD or rotating media, in
12997 order to optimize the read-ahead scheme. Previously, it was
12998 only capable of detecting this on traditional file systems
13001 * In udev, additional device properties are now read from the
13002 IAB in addition to the OUI database. Also, Bluetooth company
13003 identities are attached to the devices as well.
13005 * In service files %U may be used as specifier that is
13006 replaced by the configured user name of the service.
13008 * nspawn may now be invoked without a controlling TTY. This
13009 makes it suitable for invocation as its own service. This
13010 may be used to set up a simple containerized server system
13011 using only core OS tools.
13013 * systemd and nspawn can now accept socket file descriptors
13014 when they are started for socket activation. This enables
13015 implementation of socket activated nspawn
13016 containers. i.e. think about autospawning an entire OS image
13017 when the first SSH or HTTP connection is received. We expect
13018 that similar functionality will also be added to libvirt-lxc
13021 * journalctl will now suppress ANSI color codes when
13022 presenting log data.
13024 * systemctl will no longer show control group information for
13025 a unit if the control group is empty anyway.
13027 * logind can now automatically suspend/hibernate/shutdown the
13030 * /etc/machine-info and hostnamed now also expose the chassis
13031 type of the system. This can be used to determine whether
13032 the local system is a laptop, desktop, handset or
13033 tablet. This information may either be configured by the
13034 user/vendor or is automatically determined from ACPI and DMI
13035 information if possible.
13037 * A number of polkit actions are now bound together with "imply"
13038 rules. This should simplify creating UIs because many actions
13039 will now authenticate similar ones as well.
13041 * Unit files learnt a new condition ConditionACPower= which
13042 may be used to conditionalize a unit depending on whether an
13043 AC power source is connected or not, of whether the system
13044 is running on battery power.
13046 * systemctl gained a new "is-failed" verb that may be used in
13047 shell scripts and suchlike to check whether a specific unit
13048 is in the "failed" state.
13050 * The EnvironmentFile= setting in unit files now supports file
13051 globbing, and can hence be used to easily read a number of
13052 environment files at once.
13054 * systemd will no longer detect and recognize specific
13055 distributions. All distribution-specific #ifdeffery has been
13056 removed, systemd is now fully generic and
13057 distribution-agnostic. Effectively, not too much is lost as
13058 a lot of the code is still accessible via explicit configure
13059 switches. However, support for some distribution specific
13060 legacy configuration file formats has been dropped. We
13061 recommend distributions to simply adopt the configuration
13062 files everybody else uses now and convert the old
13063 configuration from packaging scripts. Most distributions
13064 already did that. If that's not possible or desirable,
13065 distributions are welcome to forward port the specific
13066 pieces of code locally from the git history.
13068 * When logging a message about a unit systemd will now always
13069 log the unit name in the message meta data.
13071 * localectl will now also discover system locale data that is
13072 not stored in locale archives, but directly unpacked.
13074 * logind will no longer unconditionally use framebuffer
13075 devices as seat masters, i.e. as devices that are required
13076 to be existing before a seat is considered preset. Instead,
13077 it will now look for all devices that are tagged as
13078 "seat-master" in udev. By default, framebuffer devices will
13079 be marked as such, but depending on local systems, other
13080 devices might be marked as well. This may be used to
13081 integrate graphics cards using closed source drivers (such
13082 as NVidia ones) more nicely into logind. Note however, that
13083 we recommend using the open source NVidia drivers instead,
13084 and no udev rules for the closed-source drivers will be
13085 shipped from us upstream.
13087 Contributions from: Adam Williamson, Alessandro Crismani, Auke
13088 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
13089 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
13090 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
13091 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
13092 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
13093 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
13094 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
13095 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
13096 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
13097 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
13098 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
13103 * udev gained support for loading additional device properties
13104 from an indexed database that is keyed by vendor/product IDs
13105 and similar device identifiers. For the beginning this
13106 "hwdb" is populated with data from the well-known PCI and
13107 USB database, but also includes PNP, ACPI and OID data. In
13108 the longer run this indexed database shall grow into
13109 becoming the one central database for non-essential
13110 userspace device metadata. Previously, data from the PCI/USB
13111 database was only attached to select devices, since the
13112 lookup was a relatively expensive operation due to O(n) time
13113 complexity (with n being the number of entries in the
13114 database). Since this is now O(1), we decided to add in this
13115 data for all devices where this is available, by
13116 default. Note that the indexed database needs to be rebuilt
13117 when new data files are installed. To achieve this you need
13118 to update your packaging scripts to invoke "udevadm hwdb
13119 --update" after installation of hwdb data files. For
13120 RPM-based distributions we introduced the new
13121 %udev_hwdb_update macro for this purpose.
13123 * The Journal gained support for the "Message Catalog", an
13124 indexed database to link up additional information with
13125 journal entries. For further details please check:
13127 https://www.freedesktop.org/wiki/Software/systemd/catalog
13129 The indexed message catalog database also needs to be
13130 rebuilt after installation of message catalog files. Use
13131 "journalctl --update-catalog" for this. For RPM-based
13132 distributions we introduced the %journal_catalog_update
13133 macro for this purpose.
13135 * The Python Journal bindings gained support for the standard
13136 Python logging framework.
13138 * The Journal API gained new functions for checking whether
13139 the underlying file system of a journal file is capable of
13140 properly reporting file change notifications, or whether
13141 applications that want to reflect journal changes "live"
13142 need to recheck journal files continuously in appropriate
13145 * It is now possible to set the "age" field for tmpfiles
13146 entries to 0, indicating that files matching this entry
13147 shall always be removed when the directories are cleaned up.
13149 * coredumpctl gained a new "gdb" verb which invokes gdb
13150 right-away on the selected coredump.
13152 * There's now support for "hybrid sleep" on kernels that
13153 support this, in addition to "suspend" and "hibernate". Use
13154 "systemctl hybrid-sleep" to make use of this.
13156 * logind's HandleSuspendKey= setting (and related settings)
13157 now gained support for a new "lock" setting to simply
13158 request the screen lock on all local sessions, instead of
13159 actually executing a suspend or hibernation.
13161 * systemd will now mount the EFI variables file system by
13164 * Socket units now gained support for configuration of the
13165 SMACK security label.
13167 * timedatectl will now output the time of the last and next
13168 daylight saving change.
13170 * We dropped support for various legacy and distro-specific
13171 concepts, such as insserv, early-boot SysV services
13172 (i.e. those for non-standard runlevels such as 'b' or 'S')
13173 or ArchLinux /etc/rc.conf support. We recommend the
13174 distributions who still need support this to either continue
13175 to maintain the necessary patches downstream, or find a
13176 different solution. (Talk to us if you have questions!)
13178 * Various systemd components will now bypass polkit checks for
13179 root and otherwise handle properly if polkit is not found to
13180 be around. This should fix most issues for polkit-less
13181 systems. Quite frankly this should have been this way since
13182 day one. It is absolutely our intention to make systemd work
13183 fine on polkit-less systems, and we consider it a bug if
13184 something does not work as it should if polkit is not around.
13186 * For embedded systems it is now possible to build udev and
13187 systemd without blkid and/or kmod support.
13189 * "systemctl switch-root" is now capable of switching root
13190 more than once. I.e. in addition to transitions from the
13191 initrd to the host OS it is now possible to transition to
13192 further OS images from the host. This is useful to implement
13193 offline updating tools.
13195 * Various other additions have been made to the RPM macros
13196 shipped with systemd. Use %udev_rules_update() after
13197 installing new udev rules files. %_udevhwdbdir,
13198 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
13199 %_sysctldir are now available which resolve to the right
13200 directories for packages to place various data files in.
13202 * journalctl gained the new --full switch (in addition to
13203 --all, to disable ellipsation for long messages.
13205 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
13206 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
13207 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
13208 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
13209 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
13210 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
13211 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
13212 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
13213 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
13217 * journalctl gained new --since= and --until= switches to
13218 filter by time. It also now supports nice filtering for
13219 units via --unit=/-u.
13221 * Type=oneshot services may use ExecReload= and do the
13224 * The journal daemon now supports time-based rotation and
13225 vacuuming, in addition to the usual disk-space based
13228 * The journal will now index the available field values for
13229 each field name. This enables clients to show pretty drop
13230 downs of available match values when filtering. The bash
13231 completion of journalctl has been updated
13232 accordingly. journalctl gained a new switch -F to list all
13233 values a certain field takes in the journal database.
13235 * More service events are now written as structured messages
13236 to the journal, and made recognizable via message IDs.
13238 * The timedated, localed and hostnamed mini-services which
13239 previously only provided support for changing time, locale
13240 and hostname settings from graphical DEs such as GNOME now
13241 also have a minimal (but very useful) text-based client
13242 utility each. This is probably the nicest way to changing
13243 these settings from the command line now, especially since
13244 it lists available options and is fully integrated with bash
13247 * There's now a new tool "systemd-coredumpctl" to list and
13248 extract coredumps from the journal.
13250 * We now install a README each in /var/log/ and
13251 /etc/rc.d/init.d explaining where the system logs and init
13252 scripts went. This hopefully should help folks who go to
13253 that dirs and look into the otherwise now empty void and
13254 scratch their heads.
13256 * When user-services are invoked (by systemd --user) the
13257 $MANAGERPID env var is set to the PID of systemd.
13259 * SIGRTMIN+24 when sent to a --user instance will now result
13260 in immediate termination of systemd.
13262 * gatewayd received numerous feature additions such as a
13263 "follow" mode, for live syncing and filtering.
13265 * browse.html now allows filtering and showing detailed
13266 information on specific entries. Keyboard navigation and
13267 mouse screen support has been added.
13269 * gatewayd/journalctl now supports HTML5/JSON
13270 Server-Sent-Events as output.
13272 * The SysV init script compatibility logic will now
13273 heuristically determine whether a script supports the
13274 "reload" verb, and only then make this available as
13275 "systemctl reload".
13277 * "systemctl status --follow" has been removed, use "journalctl
13280 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
13281 have been removed since they are hardly useful to be
13284 * And I'd like to take the opportunity to specifically mention
13285 Zbigniew for his great contributions. Zbigniew, you rock!
13287 Contributions from: Andrew Eikum, Christian Hesse, Colin
13288 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
13289 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
13290 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
13291 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
13292 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
13293 Jędrzejewski-Szmek, Сковорода Никита Андреевич
13297 * If /etc/vconsole.conf is non-existent or empty we will no
13298 longer load any console font or key map at boot by
13299 default. Instead the kernel defaults will be left
13300 intact. This is definitely the right thing to do, as no
13301 configuration should mean no configuration, and hard-coding
13302 font names that are different on all archs is probably a bad
13303 idea. Also, the kernel default key map and font should be
13304 good enough for most cases anyway, and mostly identical to
13305 the userspace fonts/key maps we previously overloaded them
13306 with. If distributions want to continue to default to a
13307 non-kernel font or key map they should ship a default
13308 /etc/vconsole.conf with the appropriate contents.
13310 Contributions from: Colin Walters, Daniel J Walsh, Dave
13311 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
13312 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
13316 * journalctl gained a new --cursor= switch to show entries
13317 starting from the specified location in the journal.
13319 * We now enforce a size limit on journal entry fields exported
13320 with "-o json" in journalctl. Fields larger than 4K will be
13321 assigned null. This can be turned off with --all.
13323 * An (optional) journal gateway daemon is now available as
13324 "systemd-journal-gatewayd.service". This service provides
13325 access to the journal via HTTP and JSON. This functionality
13326 will be used to implement live log synchronization in both
13327 pull and push modes, but has various other users too, such
13328 as easy log access for debugging of embedded devices. Right
13329 now it is already useful to retrieve the journal via HTTP:
13331 # systemctl start systemd-journal-gatewayd.service
13332 # wget http://localhost:19531/entries
13334 This will download the journal contents in a
13335 /var/log/messages compatible format. The same as JSON:
13337 # curl -H"Accept: application/json" http://localhost:19531/entries
13339 This service is also accessible via a web browser where a
13340 single static HTML5 app is served that uses the JSON logic
13341 to enable the user to do some basic browsing of the
13342 journal. This will be extended later on. Here's an example
13343 screenshot of this app in its current state:
13345 http://0pointer.de/public/journal-gatewayd
13347 Contributions from: Kay Sievers, Lennart Poettering, Robert
13348 Milasan, Tom Gundersen
13352 * The bash completion logic is now available for journalctl
13355 * We do not mount the "cpuset" controller anymore together with
13356 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
13357 started if no parameters are assigned to it. "cpuset" hence
13358 broke code that assumed it could create "cpu" groups and
13361 * journalctl -f will now subscribe to terminal size changes,
13362 and line break accordingly.
13364 Contributions from: Dave Reisner, Kay Sievers, Lennart
13365 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
13369 * nspawn will now create a symlink /etc/localtime in the
13370 container environment, copying the host's timezone
13371 setting. Previously this has been done via a bind mount, but
13372 since symlinks cannot be bind mounted this has now been
13373 changed to create/update the appropriate symlink.
13375 * journalctl -n's line number argument is now optional, and
13376 will default to 10 if omitted.
13378 * journald will now log the maximum size the journal files may
13379 take up on disk. This is particularly useful if the default
13380 built-in logic of determining this parameter from the file
13381 system size is used. Use "systemctl status
13382 systemd-journald.service" to see this information.
13384 * The multi-seat X wrapper tool has been stripped down. As X
13385 is now capable of enumerating graphics devices via udev in a
13386 seat-aware way the wrapper is not strictly necessary
13387 anymore. A stripped down temporary stop-gap is still shipped
13388 until the upstream display managers have been updated to
13389 fully support the new X logic. Expect this wrapper to be
13390 removed entirely in one of the next releases.
13392 * HandleSleepKey= in logind.conf has been split up into
13393 HandleSuspendKey= and HandleHibernateKey=. The old setting
13394 is not available anymore. X11 and the kernel are
13395 distinguishing between these keys and we should too. This
13396 also means the inhibition lock for these keys has been split
13399 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
13400 Poettering, Lukas Nykryn, Václav Pavlín
13404 * Whenever a unit changes state we will now log this to the
13405 journal and show along the unit's own log output in
13406 "systemctl status".
13408 * ConditionPathIsMountPoint= can now properly detect bind
13409 mount points too. (Previously, a bind mount of one file
13410 system to another place in the same file system could not be
13411 detected as mount, since they shared struct stat's st_dev
13414 * We will now mount the cgroup controllers cpu, cpuacct,
13415 cpuset and the controllers net_cls, net_prio together by
13418 * nspawn containers will now have a virtualized boot
13419 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
13420 over with a randomized ID at container initialization). This
13421 has the effect of making "journalctl -b" do the right thing
13424 * The JSON output journal serialization has been updated not
13425 to generate "endless" list objects anymore, but rather one
13426 JSON object per line. This is more in line how most JSON
13427 parsers expect JSON objects. The new output mode
13428 "json-pretty" has been added to provide similar output, but
13429 neatly aligned for readability by humans.
13431 * We dropped all explicit sync() invocations in the shutdown
13432 code. The kernel does this implicitly anyway in the kernel
13433 reboot() syscall. halt(8)'s -n option is now a compatibility
13436 * We now support virtualized reboot() in containers, as
13437 supported by newer kernels. We will fall back to exit() if
13438 CAP_SYS_REBOOT is not available to the container. Also,
13439 nspawn makes use of this now and will actually reboot the
13440 container if the containerized OS asks for that.
13442 * journalctl will only show local log output by default
13443 now. Use --merge (-m) to show remote log output, too.
13445 * libsystemd-journal gained the new sd_journal_get_usage()
13446 call to determine the current disk usage of all journal
13447 files. This is exposed in the new "journalctl --disk-usage"
13450 * journald gained a new configuration setting SplitMode= in
13451 journald.conf which may be used to control how user journals
13452 are split off. See journald.conf(5) for details.
13454 * A new condition type ConditionFileNotEmpty= has been added.
13456 * tmpfiles' "w" lines now support file globbing, to write
13457 multiple files at once.
13459 * We added Python bindings for the journal submission
13460 APIs. More Python APIs for a number of selected APIs will
13461 likely follow. Note that we intend to add native bindings
13462 only for the Python language, as we consider it common
13463 enough to deserve bindings shipped within systemd. There are
13464 various projects outside of systemd that provide bindings
13465 for languages such as PHP or Lua.
13467 * Many conditions will now resolve specifiers such as %i. In
13468 addition, PathChanged= and related directives of .path units
13469 now support specifiers as well.
13471 * There's now a new RPM macro definition for the system preset
13474 * journald will now warn if it ca not forward a message to the
13475 syslog daemon because its socket is full.
13477 * timedated will no longer write or process /etc/timezone,
13478 except on Debian. As we do not support late mounted /usr
13479 anymore /etc/localtime always being a symlink is now safe,
13480 and hence the information in /etc/timezone is not necessary
13483 * logind will now always reserve one VT for a text getty (VT6
13484 by default). Previously if more than 6 X sessions where
13485 started they took up all the VTs with auto-spawned gettys,
13486 so that no text gettys were available anymore.
13488 * udev will now automatically inform the btrfs kernel logic
13489 about btrfs RAID components showing up. This should make
13490 simple hotplug based btrfs RAID assembly work.
13492 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
13493 (but not for its children which will stay at the kernel
13494 default). This should allow setups with a lot more listening
13497 * systemd will now always pass the configured timezone to the
13498 kernel at boot. timedated will do the same when the timezone
13501 * logind's inhibition logic has been updated. By default,
13502 logind will now handle the lid switch, the power and sleep
13503 keys all the time, even in graphical sessions. If DEs want
13504 to handle these events on their own they should take the new
13505 handle-power-key, handle-sleep-key and handle-lid-switch
13506 inhibitors during their runtime. A simple way to achieve
13507 that is to invoke the DE wrapped in an invocation of:
13509 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
13511 * Access to unit operations is now checked via SELinux taking
13512 the unit file label and client process label into account.
13514 * systemd will now notify the administrator in the journal
13515 when he over-mounts a non-empty directory.
13517 * There are new specifiers that are resolved in unit files,
13518 for the hostname (%H), the machine ID (%m) and the boot ID
13521 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
13522 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
13523 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
13524 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
13525 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
13526 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
13527 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
13531 * Support for reading structured kernel messages from
13532 /dev/kmsg has now been added and is enabled by default.
13534 * Support for reading kernel messages from /proc/kmsg has now
13535 been removed. If you want kernel messages in the journal
13536 make sure to run a recent kernel (>= 3.5) that supports
13537 reading structured messages from /dev/kmsg (see
13538 above). /proc/kmsg is now exclusive property of classic
13539 syslog daemons again.
13541 * The libudev API gained the new
13542 udev_device_new_from_device_id() call.
13544 * The logic for file system namespace (ReadOnlyDirectory=,
13545 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
13546 require pivot_root() anymore. This means fewer temporary
13547 directories are created below /tmp for this feature.
13549 * nspawn containers will now see and receive all submounts
13550 made on the host OS below the root file system of the
13553 * Forward Secure Sealing is now supported for Journal files,
13554 which provide cryptographical sealing of journal files so
13555 that attackers cannot alter log history anymore without this
13556 being detectable. Lennart will soon post a blog story about
13557 this explaining it in more detail.
13559 * There are two new service settings RestartPreventExitStatus=
13560 and SuccessExitStatus= which allow configuration of exit
13561 status (exit code or signal) which will be excepted from the
13562 restart logic, resp. consider successful.
13564 * journalctl gained the new --verify switch that can be used
13565 to check the integrity of the structure of journal files and
13566 (if Forward Secure Sealing is enabled) the contents of
13569 * nspawn containers will now be run with /dev/stdin, /dev/fd/
13570 and similar symlinks pre-created. This makes running shells
13571 as container init process a lot more fun.
13573 * The fstab support can now handle PARTUUID= and PARTLABEL=
13576 * A new ConditionHost= condition has been added to match
13577 against the hostname (with globs) and machine ID. This is
13578 useful for clusters where a single OS image is used to
13579 provision a large number of hosts which shall run slightly
13580 different sets of services.
13582 * Services which hit the restart limit will now be placed in a
13585 Contributions from: Bertram Poettering, Dave Reisner, Huang
13586 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
13587 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
13591 * When running in --user mode systemd will now become a
13592 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
13593 tree a lot more organized.
13595 * A new PartOf= unit dependency type has been introduced that
13596 may be used to group services in a natural way.
13598 * "systemctl enable" may now be used to enable instances of
13601 * journalctl now prints error log levels in red, and
13602 warning/notice log levels in bright white. It also supports
13603 filtering by log level now.
13605 * cgtop gained a new -n switch (similar to top), to configure
13606 the maximum number of iterations to run for. It also gained
13607 -b, to run in batch mode (accepting no input).
13609 * The suffix ".service" may now be omitted on most systemctl
13610 command lines involving service unit names.
13612 * There's a new bus call in logind to lock all sessions, as
13613 well as a loginctl verb for it "lock-sessions".
13615 * libsystemd-logind.so gained a new call sd_journal_perror()
13616 that works similar to libc perror() but logs to the journal
13617 and encodes structured information about the error number.
13619 * /etc/crypttab entries now understand the new keyfile-size=
13622 * shutdown(8) now can send a (configurable) wall message when
13623 a shutdown is cancelled.
13625 * The mount propagation mode for the root file system will now
13626 default to "shared", which is useful to make containers work
13627 nicely out-of-the-box so that they receive new mounts from
13628 the host. This can be undone locally by running "mount
13629 --make-rprivate /" if needed.
13631 * The prefdm.service file has been removed. Distributions
13632 should maintain this unit downstream if they intend to keep
13633 it around. However, we recommend writing normal unit files
13634 for display managers instead.
13636 * Since systemd is a crucial part of the OS we will now
13637 default to a number of compiler switches that improve
13638 security (hardening) such as read-only relocations, stack
13639 protection, and suchlike.
13641 * The TimeoutSec= setting for services is now split into
13642 TimeoutStartSec= and TimeoutStopSec= to allow configuration
13643 of individual time outs for the start and the stop phase of
13646 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
13647 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
13648 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
13649 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
13650 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
13651 Gundersen, Zbigniew Jędrzejewski-Szmek
13655 * The journal and id128 C APIs are now fully documented as man
13658 * Extra safety checks have been added when transitioning from
13659 the initial RAM disk to the main system to avoid accidental
13662 * /etc/crypttab entries now understand the new keyfile-offset=
13665 * systemctl -t can now be used to filter by unit load state.
13667 * The journal C API gained the new sd_journal_wait() call to
13668 make writing synchronous journal clients easier.
13670 * journalctl gained the new -D switch to show journals from a
13671 specific directory.
13673 * journalctl now displays a special marker between log
13674 messages of two different boots.
13676 * The journal is now explicitly flushed to /var via a service
13677 systemd-journal-flush.service, rather than implicitly simply
13678 by seeing /var/log/journal to be writable.
13680 * journalctl (and the journal C APIs) can now match for much
13681 more complex expressions, with alternatives and
13684 * When transitioning from the initial RAM disk to the main
13685 system we will now kill all processes in a killing spree to
13686 ensure no processes stay around by accident.
13688 * Three new specifiers may be used in unit files: %u, %h, %s
13689 resolve to the user name, user home directory resp. user
13690 shell. This is useful for running systemd user instances.
13692 * We now automatically rotate journal files if their data
13693 object hash table gets a fill level > 75%. We also size the
13694 hash table based on the configured maximum file size. This
13695 together should lower hash collisions drastically and thus
13696 speed things up a bit.
13698 * journalctl gained the new "--header" switch to introspect
13699 header data of journal files.
13701 * A new setting SystemCallFilters= has been added to services which may
13702 be used to apply deny lists or allow lists to system calls. This is
13703 based on SECCOMP Mode 2 of Linux 3.5.
13705 * nspawn gained a new --link-journal= switch (and quicker: -j)
13706 to link the container journal with the host. This makes it
13707 very easy to centralize log viewing on the host for all
13708 guests while still keeping the journal files separated.
13710 * Many bugfixes and optimizations
13712 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
13713 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
13714 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
13719 * Several tools now understand kernel command line arguments,
13720 which are only read when run in an initial RAM disk. They
13721 usually follow closely their normal counterparts, but are
13724 * There's a new tool to analyze the readahead files that are
13725 automatically generated at boot. Use:
13727 /usr/lib/systemd/systemd-readahead analyze /.readahead
13729 * We now provide an early debug shell on tty9 if this enabled. Use:
13731 systemctl enable debug-shell.service
13733 * All plymouth related units have been moved into the Plymouth
13734 package. Please make sure to upgrade your Plymouth version
13737 * systemd-tmpfiles now supports getting passed the basename of
13738 a configuration file only, in which case it will look for it
13739 in all appropriate directories automatically.
13741 * udevadm info now takes a /dev or /sys path as argument, and
13742 does the right thing. Example:
13744 udevadm info /dev/sda
13745 udevadm info /sys/class/block/sda
13747 * systemctl now prints a warning if a unit is stopped but a
13748 unit that might trigger it continues to run. Example: a
13749 service is stopped but the socket that activates it is left
13752 * "systemctl status" will now mention if the log output was
13753 shortened due to rotation since a service has been started.
13755 * The journal API now exposes functions to determine the
13756 "cutoff" times due to rotation.
13758 * journald now understands SIGUSR1 and SIGUSR2 for triggering
13759 immediately flushing of runtime logs to /var if possible,
13760 resp. for triggering immediate rotation of the journal
13763 * It is now considered an error if a service is attempted to
13764 be stopped that is not loaded.
13766 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
13768 * systemd-analyze now supports Python 3
13770 * tmpfiles now supports cleaning up directories via aging
13771 where the first level dirs are always kept around but
13772 directories beneath it automatically aged. This is enabled
13773 by prefixing the age field with '~'.
13775 * Seat objects now expose CanGraphical, CanTTY properties
13776 which is required to deal with very fast bootups where the
13777 display manager might be running before the graphics drivers
13778 completed initialization.
13780 * Seat objects now expose a State property.
13782 * We now include RPM macros for service enabling/disabling
13783 based on the preset logic. We recommend RPM based
13784 distributions to make use of these macros if possible. This
13785 makes it simpler to reuse RPM spec files across
13788 * We now make sure that the collected systemd unit name is
13789 always valid when services log to the journal via
13792 * There's a new man page kernel-command-line(7) detailing all
13793 command line options we understand.
13795 * The fstab generator may now be disabled at boot by passing
13796 fstab=0 on the kernel command line.
13798 * A new kernel command line option modules-load= is now understood
13799 to load a specific kernel module statically, early at boot.
13801 * Unit names specified on the systemctl command line are now
13802 automatically escaped as needed. Also, if file system or
13803 device paths are specified they are automatically turned
13804 into the appropriate mount or device unit names. Example:
13806 systemctl status /home
13807 systemctl status /dev/sda
13809 * The SysVConsole= configuration option has been removed from
13810 system.conf parsing.
13812 * The SysV search path is no longer exported on the D-Bus
13815 * The Names= option has been removed from unit file parsing.
13817 * There's a new man page bootup(7) detailing the boot process.
13819 * Every unit and every generator we ship with systemd now
13820 comes with full documentation. The self-explanatory boot is
13823 * A couple of services gained "systemd-" prefixes in their
13824 name if they wrap systemd code, rather than only external
13825 code. Among them fsck@.service which is now
13826 systemd-fsck@.service.
13828 * The HaveWatchdog property has been removed from the D-Bus
13831 * systemd.confirm_spawn= on the kernel command line should now
13834 * There's a new man page crypttab(5) which details all options
13835 we actually understand.
13837 * systemd-nspawn gained a new --capability= switch to pass
13838 additional capabilities to the container.
13840 * timedated will now read known NTP implementation unit names
13841 from /usr/lib/systemd/ntp-units.d/*.list,
13842 systemd-timedated-ntp.target has been removed.
13844 * journalctl gained a new switch "-b" that lists log data of
13845 the current boot only.
13847 * The notify socket is in the abstract namespace again, in
13848 order to support daemons which chroot() at start-up.
13850 * There is a new Storage= configuration option for journald
13851 which allows configuration of where log data should go. This
13852 also provides a way to disable journal logging entirely, so
13853 that data collected is only forwarded to the console, the
13854 kernel log buffer or another syslog implementation.
13856 * Many bugfixes and optimizations
13858 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
13859 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
13860 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
13861 Shawn Landden, Tom Gundersen
13865 * "systemctl help <unit>" now shows the man page if one is
13868 * Several new man pages have been added.
13870 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
13871 MaxLevelConsole= can now be specified in
13872 journald.conf. These options allow reducing the amount of
13873 data stored on disk or forwarded by the log level.
13875 * TimerSlackNSec= can now be specified in system.conf for
13876 PID1. This allows system-wide power savings.
13878 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
13879 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
13884 * logind is now capable of (optionally) handling power and
13885 sleep keys as well as the lid switch.
13887 * journalctl now understands the syntax "journalctl
13888 /usr/bin/avahi-daemon" to get all log output of a specific
13891 * CapabilityBoundingSet= in system.conf now also influences
13892 the capability bound set of usermode helpers of the kernel.
13894 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
13895 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
13896 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
13897 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
13901 * Note that we skipped 139 releases here in order to set the
13902 new version to something that is greater than both udev's
13903 and systemd's most recent version number.
13905 * udev: all udev sources are merged into the systemd source tree now.
13906 All future udev development will happen in the systemd tree. It
13907 is still fully supported to use the udev daemon and tools without
13908 systemd running, like in initramfs or other init systems. Building
13909 udev though, will require the *build* of the systemd tree, but
13910 udev can be properly *run* without systemd.
13912 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
13913 should be used to create dead device nodes as workarounds for broken
13916 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
13917 no longer supported. udev_monitor_new_from_netlink() needs to be
13918 used to subscribe to events.
13920 * udev: when udevd is started by systemd, processes which are left
13921 behind by forking them off of udev rules, are unconditionally cleaned
13922 up and killed now after the event handling has finished. Services or
13923 daemons must be started as systemd services. Services can be
13924 pulled-in by udev to get started, but they can no longer be directly
13925 forked by udev rules.
13927 * udev: the daemon binary is called systemd-udevd now and installed
13928 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
13929 to adapt to that, create symlink, or rename the binary after building
13932 * libudev no longer provides these symbols:
13933 udev_monitor_from_socket()
13934 udev_queue_get_failed_list_entry()
13935 udev_get_{dev,sys,run}_path()
13936 The versions number was bumped and symbol versioning introduced.
13938 * systemd-loginctl and systemd-journalctl have been renamed
13939 to loginctl and journalctl to match systemctl.
13941 * The config files: /etc/systemd/systemd-logind.conf and
13942 /etc/systemd/systemd-journald.conf have been renamed to
13943 logind.conf and journald.conf. Package updates should rename
13944 the files to the new names on upgrade.
13946 * For almost all files the license is now LGPL2.1+, changed
13947 from the previous GPL2.0+. Exceptions are some minor stuff
13948 of udev (which will be changed to LGPL2.1 eventually, too),
13949 and the MIT licensed sd-daemon.[ch] library that is suitable
13950 to be used as drop-in files.
13952 * systemd and logind now handle system sleep states, in
13953 particular suspending and hibernating.
13955 * logind now implements a sleep/shutdown/idle inhibiting logic
13956 suitable for a variety of uses. Soonishly Lennart will blog
13957 about this in more detail.
13959 * var-run.mount and var-lock.mount are no longer provided
13960 (which previously bind mounted these directories to their new
13961 places). Distributions which have not converted these
13962 directories to symlinks should consider stealing these files
13963 from git history and add them downstream.
13965 * We introduced the Documentation= field for units and added
13966 this to all our shipped units. This is useful to make it
13967 easier to explore the boot and the purpose of the various
13970 * All smaller setup units (such as
13971 systemd-vconsole-setup.service) now detect properly if they
13972 are run in a container and are skipped when
13973 appropriate. This guarantees an entirely noise-free boot in
13974 Linux container environments such as systemd-nspawn.
13976 * A framework for implementing offline system updates is now
13977 integrated, for details see:
13978 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
13980 * A new service type Type=idle is available now which helps us
13981 avoiding ugly interleaving of getty output and boot status
13984 * There's now a system-wide CapabilityBoundingSet= option to
13985 globally reduce the set of capabilities for the
13986 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
13987 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
13988 even CAP_NET_ADMIN system-wide for secure systems.
13990 * There are now system-wide DefaultLimitXXX= options to
13991 globally change the defaults of the various resource limits
13992 for all units started by PID 1.
13994 * Harald Hoyer's systemd test suite has been integrated into
13995 systemd which allows easy testing of systemd builds in qemu
13996 and nspawn. (This is really awesome! Ask us for details!)
13998 * The fstab parser is now implemented as generator, not inside
14001 * systemctl will now warn you if .mount units generated from
14002 /etc/fstab are out of date due to changes in fstab that
14003 have not been read by systemd yet.
14005 * systemd is now suitable for usage in initrds. Dracut has
14006 already been updated to make use of this. With this in place
14007 initrds get a slight bit faster but primarily are much
14008 easier to introspect and debug since "systemctl status" in
14009 the host system can be used to introspect initrd services,
14010 and the journal from the initrd is kept around too.
14012 * systemd-delta has been added, a tool to explore differences
14013 between user/admin configuration and vendor defaults.
14015 * PrivateTmp= now affects both /tmp and /var/tmp.
14017 * Boot time status messages are now much prettier and feature
14018 proper english language. Booting up systemd has never been
14021 * Read-ahead pack files now include the inode number of all
14022 files to pre-cache. When the inode changes the pre-caching
14023 is not attempted. This should be nicer to deal with updated
14024 packages which might result in changes of read-ahead
14027 * We now temporaritly lower the kernel's read_ahead_kb variable
14028 when collecting read-ahead data to ensure the kernel's
14029 built-in read-ahead does not add noise to our measurements
14030 of necessary blocks to pre-cache.
14032 * There's now RequiresMountsFor= to add automatic dependencies
14033 for all mounts necessary for a specific file system path.
14035 * MountAuto= and SwapAuto= have been removed from
14036 system.conf. Mounting file systems at boot has to take place
14039 * nspawn now learned a new switch --uuid= to set the machine
14040 ID on the command line.
14042 * nspawn now learned the -b switch to automatically search
14043 for an init system.
14045 * vt102 is now the default TERM for serial TTYs, upgraded from
14048 * systemd-logind now works on VT-less systems.
14050 * The build tree has been reorganized. The individual
14051 components now have directories of their own.
14053 * A new condition type ConditionPathIsReadWrite= is now available.
14055 * nspawn learned the new -C switch to create cgroups for the
14056 container in other hierarchies.
14058 * We now have support for hardware watchdogs, configurable in
14061 * The scheduled shutdown logic now has a public API.
14063 * We now mount /tmp as tmpfs by default, but this can be
14064 masked and /etc/fstab can override it.
14066 * Since udisks does not make use of /media anymore we are not
14067 mounting a tmpfs on it anymore.
14069 * journalctl gained a new --local switch to only interleave
14070 locally generated journal files.
14072 * We can now load the IMA policy at boot automatically.
14074 * The GTK tools have been split off into a systemd-ui.
14076 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
14077 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
14078 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
14079 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
14080 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
14081 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
14082 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
14083 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
14084 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
14089 * This is mostly a bugfix release
14091 * Support optional initialization of the machine ID from the
14092 KVM or container configured UUID.
14094 * Support immediate reboots with "systemctl reboot -ff"
14096 * Show /etc/os-release data in systemd-analyze output
14098 * Many bugfixes for the journal, including endianness fixes and
14099 ensuring that disk space enforcement works
14101 * sd-login.h is C++ compatible again
14103 * Extend the /etc/os-release format on request of the Debian
14106 * We now refuse non-UTF8 strings used in various configuration
14107 and unit files. This is done to ensure we do not pass invalid
14108 data over D-Bus or expose it elsewhere.
14110 * Register Mimo USB Screens as suitable for automatic seat
14113 * Read SELinux client context from journal clients in a race
14116 * Reorder configuration file lookup order. /etc now always
14117 overrides /run in order to allow the administrator to always
14118 and unconditionally override vendor-supplied or
14119 automatically generated data.
14121 * The various user visible bits of the journal now have man
14122 pages. We still lack man pages for the journal API calls
14125 * We now ship all man pages in HTML format again in the
14128 Contributions from: Dave Reisner, Dirk Eibach, Frederic
14129 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
14130 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
14135 * This is mostly a bugfix release
14137 * systems lacking /etc/os-release are no longer supported.
14139 * Various functionality updates to libsystemd-login.so
14141 * Track class of PAM logins to distinguish greeters from
14142 normal user logins.
14144 Contributions from: Kay Sievers, Lennart Poettering, Michael
14149 * This is an important bugfix release for v41.
14151 * Building man pages is now optional which should be useful
14152 for those building systemd from git but unwilling to install
14155 * Watchdog support for supervising services is now usable. In
14156 a future release support for hardware watchdogs
14157 (i.e. /dev/watchdog) will be added building on this.
14159 * Service start rate limiting is now configurable and can be
14160 turned off per service. When a start rate limit is hit a
14161 reboot can automatically be triggered.
14163 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
14165 Contributions from: Benjamin Franzke, Bill Nottingham,
14166 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
14167 Schmidt, Michał Górny, Piotr Drąg
14171 * The systemd binary is installed /usr/lib/systemd/systemd now;
14172 An existing /sbin/init symlink needs to be adapted with the
14175 * The code that loads kernel modules has been ported to invoke
14176 libkmod directly, instead of modprobe. This means we do not
14177 support systems with module-init-tools anymore.
14179 * Watchdog support is now already useful, but still not
14182 * A new kernel command line option systemd.setenv= is
14183 understood to set system wide environment variables
14184 dynamically at boot.
14186 * We now limit the set of capabilities of systemd-journald.
14188 * We now set SIGPIPE to ignore by default, since it only is
14189 useful in shell pipelines, and has little use in general
14190 code. This can be disabled with IgnoreSIPIPE=no in unit
14193 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
14194 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
14199 * This is mostly a bugfix release
14201 * We now expose the reason why a service failed in the
14202 "Result" D-Bus property.
14204 * Rudimentary service watchdog support (will be completed over
14205 the next few releases.)
14207 * When systemd forks off in order execute some service we will
14208 now immediately changes its argv[0] to reflect which process
14209 it will execute. This is useful to minimize the time window
14210 with a generic argv[0], which makes bootcharts more useful
14212 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
14213 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
14214 Mike Kazantsev, Ray Strode
14218 * This is mostly a test release, but incorporates many
14221 * New systemd-cgtop tool to show control groups by their
14224 * Linking against libacl for ACLs is optional again. If
14225 disabled, support tracking device access for active logins
14226 goes becomes unavailable, and so does access to the user
14227 journals by the respective users.
14229 * If a group "adm" exists, journal files are automatically
14230 owned by them, thus allow members of this group full access
14231 to the system journal as well as all user journals.
14233 * The journal now stores the SELinux context of the logging
14234 client for all entries.
14236 * Add C++ inclusion guards to all public headers
14238 * New output mode "cat" in the journal to print only text
14239 messages, without any meta data like date or time.
14241 * Include tiny X server wrapper as a temporary stop-gap to
14242 teach XOrg udev display enumeration. This is used by display
14243 managers such as gdm, and will go away as soon as XOrg
14244 learned native udev hotplugging for display devices.
14246 * Add new systemd-cat tool for executing arbitrary programs
14247 with STDERR/STDOUT connected to the journal. Can also act as
14248 BSD logger replacement, and does so by default.
14250 * Optionally store all locally generated coredumps in the
14251 journal along with meta data.
14253 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
14254 writing short strings to files (for usage for /sys), and for
14255 creating symlinks, character and block device nodes.
14257 * New unit file option ControlGroupPersistent= to make cgroups
14258 persistent, following the mechanisms outlined in
14259 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
14261 * Support multiple local RTCs in a sane way
14263 * No longer monopolize IO when replaying readahead data on
14264 rotating disks, since we might starve non-file-system IO to
14265 death, since fanotify() will not see accesses done by blkid,
14268 * Do not show kernel threads in systemd-cgls anymore, unless
14269 requested with new -k switch.
14271 Contributions from: Dan Horák, Kay Sievers, Lennart
14272 Poettering, Michal Schmidt
14276 * This is mostly a test release, but incorporates many
14279 * The git repository moved to:
14280 git://anongit.freedesktop.org/systemd/systemd
14281 ssh://git.freedesktop.org/git/systemd/systemd
14283 * First release with the journal
14284 http://0pointer.de/blog/projects/the-journal.html
14286 * The journal replaces both systemd-kmsg-syslogd and
14287 systemd-stdout-bridge.
14289 * New sd_pid_get_unit() API call in libsystemd-logind
14291 * Many systemadm clean-ups
14293 * Introduce remote-fs-pre.target which is ordered before all
14294 remote mounts and may be used to start services before all
14297 * Added Mageia support
14299 * Add bash completion for systemd-loginctl
14301 * Actively monitor PID file creation for daemons which exit in
14302 the parent process before having finished writing the PID
14303 file in the daemon process. Daemons which do this need to be
14304 fixed (i.e. PID file creation must have finished before the
14305 parent exits), but we now react a bit more gracefully to them.
14307 * Add colourful boot output, mimicking the well-known output
14308 of existing distributions.
14310 * New option PassCredentials= for socket units, for
14311 compatibility with a recent kernel ABI breakage.
14313 * /etc/rc.local is now hooked in via a generator binary, and
14314 thus will no longer act as synchronization point during
14317 * systemctl list-unit-files now supports --root=.
14319 * systemd-tmpfiles now understands two new commands: z, Z for
14320 relabelling files according to the SELinux database. This is
14321 useful to apply SELinux labels to specific files in /sys,
14322 among other things.
14324 * Output of SysV services is now forwarded to both the console
14325 and the journal by default, not only just the console.
14327 * New man pages for all APIs from libsystemd-login.
14329 * The build tree got reorganized and the build system is a
14330 lot more modular allowing embedded setups to specifically
14331 select the components of systemd they are interested in.
14333 * Support for Linux systems lacking the kernel VT subsystem is
14336 * configure's --with-rootdir= got renamed to
14337 --with-rootprefix= to follow the naming used by udev and
14340 * Unless specified otherwise we will now install to /usr instead
14341 of /usr/local by default.
14343 * Processes with '@' in argv[0][0] are now excluded from the
14344 final shut-down killing spree, following the logic explained
14346 https://systemd.io/ROOT_STORAGE_DAEMONS/
14348 * All processes remaining in a service cgroup when we enter
14349 the START or START_PRE states are now killed with
14350 SIGKILL. That means it is no longer possible to spawn
14351 background processes from ExecStart= lines (which was never
14352 supported anyway, and bad style).
14354 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
14355 reloading of units together.
14357 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
14358 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
14359 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
14360 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
14361 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek