1 systemd System and Service Manager
3 CHANGES WITH 258 in spe:
7 * Support for cgroup v1 ('legacy' and 'hybrid' hierarchies) has been
8 removed. cgroup v2 ('unified' hierarchy) will always be mounted
9 during system bootup and systemd-nspawn container initialization.
11 * The minimum kernel baseline version has been bumped to v5.4 (released
12 in 2019), with the recommended version now going up to v5.7. Consult
13 the README file for a list of required kernel APIs.
15 * The default access mode of tty/pts device nodes has been changed to
16 0600, which was 0620 in the older releases, due to general security
17 concerns about terminals being written to by other users. To restore
18 the old default access mode, use the '-Dtty-mode=0620' meson build
19 option. (This effectively means "mesg n" is now the default, rather
20 than "mesg y", see mesg(1) man page for help.)
22 * systemd-run's --expand-environment= switch, which was disabled
23 by default when combined with --scope, has been changed to be
24 enabled by default. This brings cmdline expansion of transient
25 scopes on par with services.
27 * systemd-logind PAM sessions that previously were automatically
28 determined to be of class "background", and which are owned by root
29 or system accounts, will now automatically be set to class
30 "background-light" instead. PAM sessions that previously were
31 automatically determined to be of class "user", and which are owned
32 by non-root system users, will now automatically be set to class
33 "user-light" instead. This effectively means that cron jobs or FTP
34 sessions (i.e. all PAM sessions that have no TTY assigned and neither
35 are graphical) for system users no longer pull in a service manager
36 by default. This behaviour can be changed by explicitly setting the
37 session class (for example via the class= parameter to
38 pam_systemd.so, or by setting the XDG_SESSION_CLASS environment
39 variable as input for the service's PAM stack). This change does not
40 affect graphical sessions, nor does it affect regular users. This is
41 an incompatible change of sorts, since per-user services will
42 typically not be available for such PAM sessions of system users.
44 * systemd-udevd ignores OWNER=/GROUP= settings with a non-system
45 user/group specified in udev rules files, to avoid device nodes being
46 owned by a non-system user/group. It is recommended to check udev
47 rules files with 'udevadm verify' and/or 'udevadm test' commands if
48 the specified user/group in OWNER=/GROUP= are valid.
49 Similarly, systemd-networkd refuses User=/Group= settings with a
50 non-system user/group specified in .netdev files for Tun/Tap
53 * systemd-cryptenroll, systemd-repart and systemd-creds no longer
54 default to locking TPM2 enrollments to the current, literal value of
55 PCR 7, i.e. the PCR the SecureBoot policy is measured into by the
56 firmware. This change reflects the fact that nowadays SecureBoot
57 policies are updated (at least) as frequently as firmware code
58 (simply because SecureBoot policy updates are typically managed by
59 fwupd these days). The new default PCR mask for new TPM2 enrollments
60 is thus empty by default. It is recommended to use managed
61 systemd-pcrlock policies for binding to PCR 7 instead (as well as
62 combining such policies with signed policies for PCR 11). Or in other
63 words, it's recommended to make more use of the logic behind the
64 --tpm2-public-key=, --tpm2-public-key-pcrs= and --tpm2-pcrlock=
65 switches of the mentioned tools in place of --tpm2-pcrs=.
67 * Support for the SystemdOptions EFI variable has been removed.
69 * The meson options 'split-usr', 'rootlibdir', 'rootprefix' (deprecated
70 in v255), 'default-hierarchy' (v256), and 'nscd' (v257) have been
73 * OpenSSL is now the only supported cryptography backend for
74 systemd-resolved and systemd-importd, and support for gnutls and
75 gcrypt has been removed. Hence, 'gnutls' for 'dns-over-tls=' meson
76 option has been deprecated. Also, the 'cryptolib' meson option has
77 been deprecated. They will be removed in a future release.
79 * systemd-logind's session tracking, which used to be performed via a
80 FIFO installed in the client, has been fully switched to be
81 pidfd-based. The fd returned by CreateSession() and related calls is
82 therefore unused. Moreover, the exit of session leader process will
83 immediately cause the session to be stopped.
85 * To work around limitations of X11's keyboard handling systemd's
86 keyboard mapping hardware database (hwdb.d/60-keyboard.hwdb) so far
87 mapped the microphone mute and touchpad on/off/toggle keys to the
88 function keys F20, F21, F22, F23 instead of their correct key codes.
89 This key code mangling has been removed from udev.
91 To maintain compatibility with X11 applications that rely on the old
92 function key code mappings, this mangling has now been added to the
93 relevant X11 keyboard driver modules. In order to ensure these keys
94 continue to work, update to xf86-input-evdev >= 2.11.0 and
95 xf86-input-libinput >= 1.5.0 before updating to systemd >= 258.
97 * The D-Bus method org.freedesktop.systemd1.StartAuxiliaryScope() has
98 been removed, which was deprecated since v257.
100 * systemd-networkd now supports configuring the timeout for IPv4
101 Duplicate Address Detection via a new setting
102 IPv4DuplicateAddressDetectionTimeoutSec=. The default timeout value
103 has been changed from 7 seconds to 200 milliseconds.
105 * systemd-networkd previously emitted the machine ID as chassis ID
106 through LLDP protocol, but now emits a deterministic ID,
107 cryptographically derived from the machine ID as chassis ID. If you
108 want to use the previous behavior, please set
109 SYSTEMD_LLDP_SEND_MACHINE_ID=1 environment variable to
112 * Support for the !! command line prefix on ExecStart= lines (and
113 related) has been removed, and if specified will be ignored. The
114 concept was supposed to provide compatibility with kernels that
115 predated the introduction of "ambient" process capabilities. However,
116 the kernel baseline of the systemd project is now far beyond any
117 kernels that lacked support for it, hence the prefix serves no
120 * The default keyring for systemd-importd and related tools, shipped
121 in /usr/lib/systemd/, has been renamed from import-pubring.gpg to
122 import-pubring.pgp, as it is supported by other PGP tools as well as
123 GPG. The local keyring /etc/systemd/import-pubring.gpg is still parsed
124 if present, to preserve backward compatibility.
126 * Normally, per-user encrypted credentials are decrypted via the the
127 systemd-creds.socket Varlink service, while the per-system ones are
128 directly encrypted within the execution context of the intended
129 service (which hence typically required access to /dev/tpmrm0). This
130 has been changed: units that enable either PrivateDevices= or use
131 DeviceAllow=/DevicePolicy= (and thus restrict access to device nodes)
132 will now also make use of the systemd-creds.socket Varlink
133 functionality, and will not attempt to decrypt the credentials
134 in-process (and attempt to try to talk to the TPM for
135 that). Previously, encrypted credentials for per-system services were
136 incompatible with PrivateDevices= and resulted in automatic extension
137 of the DeviceAllow= list. The latter behaviour has been removed.
139 * The command 'journalctl --follow' now exits with success on
140 SIGTERM/SIGINT, or its pipe STDOUT is disconnected.
142 Announcements of Future Feature Removals:
144 * Support for System V service scripts is deprecated and will be
145 removed in v259. Please make sure to update your software
146 *now* to include a native systemd unit file instead of a legacy
147 System V script to retain compatibility with future systemd releases.
149 * Support for systemd-repart's FactoryReset EFI variable has been
150 deprecated and support for it will be removed in v260. Use the newer,
151 more generic FactoryResetRequest variable instead, which can be
152 managed by "systemd-factory-reset request" and "systemd-factory-reset
155 * The meson option 'integration-tests' has been deprecated, and will be
156 removed in a future release.
158 Service manager/PID1:
160 * The PrivateUsers= unit setting now accepts a new value "full", which
161 is similar to "identity", but maps the whole 32bit UID range instead
162 of just the first 2¹⁶.
164 * The ProtectHostname= unit setting now accepts a new value "private",
165 which is similar to "yes", but allows the unit's processes to
166 modify the hostname. Since a UTC namespace is allocated for the unit
167 this hostname change remains local to the unit, and does not affect
168 the system as a whole. Optionally, the "private" string may be
169 suffixed by a colon and a literal hostname specification, which is
170 then used to initialize the hostname of the namespace to.
172 * .mount units now also support systemd credentials
173 (i.e. SetCredential=/LoadCredential=/ImportCredential= and related
174 settings). Previously this was available for service units only.
176 * A new unit file condition ConditionKernelModuleLoaded= has been added
177 that may be used to check if a certain kernel module is already
178 loaded (or built into the kernel). This is used to shortcut
179 modprobe@.service instances, reducing redundant explicit modprobe
180 invocations at boot to cover for kernels that have various subsystems
181 built-in, while still providing support for kernels that have those
182 subsystems built as loadable modules.
184 * Encrypted systemd service credentials are now available for user
185 services too, including if locked to TPM. Previously, they could only
186 be used for system services.
188 * Services instantiated for Accept=yes socket units will now include
189 the Linux socket cookie (SO_COOKIE) in the instance name, as well as
190 the PIDFD inode ID for the peer (the latter is only available for
191 AF_UNIX sockets). This should make it easier to match specific
192 service instances to the connections and peers they are associated
195 * The security rules enforced by the per-unit AttachProcesses() bus API
196 call have been relaxed a bit: unprivileged clients may now use the
197 call on arbitrary processes which run in any user namespace owned by
198 the client's UID. Previously, a stricter rule applied, that required
199 the UID of the process to move and of the client to match exactly.
201 * A new per-unit RemoveSubgroup() D-Bus API call has been added that
202 makes the service manager attempt to remove a sub-cgroup of units
203 with cgroup delegation enabled. This is useful for unprivileged user
204 namespace operation, where subgroups might be owned by user IDs that
205 do not match the user ID the unit was delegated to, as is typical in
206 user namespace scenarios. Per-user service managers will use this new
207 call provided by the per-system service manager to clean up user
208 units that contain cgroups owned by user namespace UIDs.
210 * .mount units gained support for a special x-systemd.graceful-option=
211 pseudo-mount option, which may be used to list additional mount
212 options that shall be used for the mount when it is established,
213 under the condition the local kernel supports them. If the local
214 kernel does not, they are automatically removed from the option
215 string. This only works for kernel-level mount options, not for those
216 implemented in userspace. This is useful for various purposes, for
217 example to include "usrquota" for tmpfs mount options where that's
220 * Per-user quota is now enabled on /dev/shm/ and /tmp/ (the latter only
223 * If PAMName= is used for a service and the PAM session prompts for a
224 password, it will now be queried via the systemd-ask-password
225 logic. Previously the prompt would simply be denied, typically causing
226 the PAM session (and thus service activation) to fail. One effect of
227 this change is that when lingering is enabled for a systemd-homed
228 user the user's password will now be prompted at boot to unlock the
229 user's home directory in order to be able to start the per-user
230 service manager early, as requested.
232 * The $MAINPID and $MANAGERPID environment variables we pass to
233 processes executed for service units are now paired with new
234 environment variables $MAINPIDFDID and $MANAGERPIDFDID. These new
235 environment variables contain the numeric inode ID of the pidfd for
236 the relevant process. As these 64bit IDs are unique for all processes
237 of a specific Linux boot they can be used to race-freely reference a
238 process, unlike the PID which is subject to races by recycling.
240 * So far the ConditionHost= condition matched against the local host
241 name and machine UUID. It now also matches against the local product
242 ID of the system (as provided by SMBIOS/DMI), and the boot ID.
244 * A new setting DelegateNamespaces= for units has been added, which
245 controls which type of Linux namespaces to delegate to the invoked
246 unit processes. This primarily controls if the listed namespace types
247 shall be owned by the host user namespace, or by the private user
248 namespace of the unit. In the former case services cannot modify the
249 relevant namespaces since they don't own it, in the latter case they
252 * If the service manager receives a RESTART_RESET=1 sd_notify() message
253 from a service, it will now reset the automatic restart counter it
254 maintains for the service. This is useful to give services control
255 over RestartMaxDelaySec=/RestartSteps= progress.
257 * The /etc/hostname file may now include question mark characters
258 ("?"), which when read will be initialized by hexadecimal digits
259 hashed from the machine ID. This is useful when managing a fleet of
260 devices that each shall have a valid and distinct hostname, generated
261 in a predictable fashion. Example: if /etc/hostname contains
262 "foobar-????-????" each booted system will end up with a hostname
263 such as "foobar-7aaf-846c" or similar.
265 * ConditionKernelVersion= has been replaced by a more generic
266 ConditionVersion= setting, that can check the versions of more key
267 components of the OS, besides the kernel. Initially, that's systemd's
268 and glibc's versions. The older setting remains supported for
271 * Slice units gained new ConcurrencySoftMax= and ConcurrencyHardMax=
272 settings which control how many concurrent units may be active and
273 queued for the slice at the same time. If more services are queued
274 for a slice than the soft limit, they won't be dispatched until the
275 concurrency falls below the limit again, but they remain in the job
276 queue. If more services are queued than the hard limit the jobs will
277 fail. This introduces a powerful job execution mechanism to systemd,
278 with strong resource management, and support for hierarchial job
279 pools (by means of slices).
281 * ExecStart= lines (and the other ExecXYZ= lines) gained a new '|'
282 special prefix that if specified causes the command line to be
285 * A basic Varlink API is now implemented in the service manager that
286 can be used to determine its current state, and list units and their
289 * Processes invoked via the .socket Accept=yes logic will now get an
290 environment variable $SO_COOKIE that contains the Linux socket
291 cookie (which otherwise can be acquired via getsockopt()) of the
292 connection socket, formatted in decimal.
294 * When a service's configuration is reloaded (via "systemctl reload" or
295 an equivalent operation), any confext images for the services are
298 * A new RandomizedOffsetSec= setting has been added to .timer units
299 which allows configured of a randomized but stable time offset for
300 when the timer shall elapse.
302 * Whenever a TTY is initialized by the service manager, an attempt is
303 made to read the terminfo identifier from it via DCS sequences, as
304 part of the regular ANSI sequence initialization scheme. The
305 identifier is used to initialize $TERM. This is not done if $TERM is
306 already set from some other sources. Note that the DCS sequence for
307 this is widely supported, but not universal (at this point VTE-based
308 terminal emulators lack the necessary support). This functionality
309 should be particularly useful on serial TTYs as $TERM information
310 will likely be initialized to a useful value instead of a badly
311 guessed default of vt220.
313 * .socket units gained a new PassPIDFD= setting that controls the new
314 SO_PASSPIDFD socket option for AF_UNIX socket. There's also a new
315 setting AcceptFileDescriptors= that controls the new SO_PASSRIGHTS.
317 * A new job type "lenient" has been added, that is similar to the
318 existing "fail" job mode, and which will fail the submitted
319 transaction immediately if it would stop any currently running unit.
321 * .socket units gained a new pair of settings DeferTrigger= and
322 DeferTriggerMaxSec= which modify triggering behaviour of the
323 socket. When used this will cause the triggered unit to be enqueued
324 with the new "lenient" job mode, and if the submission of the
325 transaction fails it is later retried to be submitted (up to a
326 configurable timeout), whenever a unit is stopped.
328 * The "preset" logic has been extended so that there are now three
329 preset directories: one that declares the default enablement state
330 for per-system services run on the host, one for per-user services,
331 and – now new – one for per-system services that are run in the
332 initrd. This reflects the fact that in many cases services that shall
333 be enabled by default on the host should not be enabled by default in
334 the initrd, or vice versa. Note that while the regular per-system
335 preset policy defaults to enabled, the one for the initrd defaults to
338 * There are now new per-service settings
339 StateDirectoryQuota=/StateDirectoryAccounting=,
340 CacheDirectoryQuota=/CacheDirectoryAccounting=,
341 LogsDirectoryQuota=/LogsDirectoryAccounting= which allow doing
342 per-unit quota of the indicated per-unit directories. This is
343 implemented via project quota, as supported by xfs and ext4. This
344 does not support btrfs, currently. If quota accounting is enabled
345 this information is shown in the usual "systemctl status" output.
347 * The service manager gained a new KillUnitSubgroup() syscall which may
348 be used to send a signal to a sub-control group of the unit's control
349 group. systemctl kill gained a new --kill-subgroup= switch to make
350 this available from the shell.
352 * A new PrivateBPF= switch has been added for unit files, which may be
353 used to mount a private bpffs instance for the unit's processes.
355 systemd-journald & journal-remote:
357 * journalctl's --setup-keys command now supports JSON output.
359 * HTTP compression negotiation has been added to journal-upload and
362 * journal-remote/journal-upload now support inserting additional HTTP
363 fields into their requests, via the Header= configuration file setting.
365 * journalctl gained a new --synchronize-on-exit=yes switch. If
366 specified in combination with --follow and the journalctl process
367 receives SIGINT (for example because the user hits Ctrl-C), a
368 synchronization request is enqueued to systemd-journald, and log
369 output continues until it completes. Or in other words, when this
370 option is used any log output submitted before the SIGINT is
371 guaranteed to be shown before journactl exits.
373 * systemd-journald's Synchronize() Varlink call has been reworked so
374 that it no longer returns only once the logging subsystem has become
375 completely idle, but already when all messages queued before the call
376 was initiated are definitely written to disk. Effectively this means
377 that the call is now guaranteed to complete in bounded time, even
378 though it's slightly weaker in effect.
380 * Many of systemd-journald's Varlink calls (such as the aforementioned
381 Synchronize()) are now available to unprivileged clients.
383 systemd-udevd & systemd-hwdb:
385 * A new udev property ID_NET_BRING_UP_BEFORE_JOINING_BRIDGE= is now
386 supported that may be set on network interface devices (via hwdb),
387 and tells systemd-networkd to bring the interface up before joining
388 it to a bridge device.
390 * A new udev property ID_NET_NAME_INCLUDE_DOMAIN= is now supported that
391 may be set on network interface devices (via hwdb), that indicates
392 that the automatic network device naming logic should suppress
393 inclusion of the PCI domain in the naming scheme. This is used for
396 * A new udev property ID_AV_LIGHTS= has been defined that may be set on
397 USB controlled A/V lights. Devices marked like this (via hwdb) will
398 have the uaccess logic enabled, i.e. they will be associated with a
399 seat and unprivileged users will get access to them.
401 * udevadm's trigger command gained a switch --include-parents. If
402 specified udevadm will not just trigger all devices matching whatever
403 is specified otherwise on the command line, but also all parent
404 devices of these devices.
406 * systemd-udevd now provides a Varlink interface with various runtime
407 and lifecycle operations. It mostly replaces the previous private,
408 undocumented "control" IPC API spoken between udevadm and
411 * .link files gained two new knobs ReceiveFCS= (which controls whether
412 to pass the Frame Check Sequence value up the stack) and ReceiveAll=
413 (which controls whether to accept damaged Ethernet frames). It also
414 gained a knob PartialGenericSegmentationOffload= for controlling
417 * 'udevadm info/trigger/test/test-builtin' commands now also take device
418 IDs to specify devices.
420 * udevadm test gained a new "--verbose" switch for generating
421 additional debug output for the test.
423 * The OPTIONS= udev expression now supports the new "dump" value, which
424 will result in the current event's status to be logged at the moment
425 the expression is processed. This is useful for debugging udev rules.
427 * A new kernel command line option udev.trace= has been added that
428 allows enabling udev's tracing logic while booting an OS. udevadm
429 control gained a new --trace= switch to change the same setting at
432 * udevadm test gained a new --extra-rules-dir= switch which may be
433 used to look for udev rules in additional directories for testing
436 * udevadm gained a new "cat" command for showing the contents of
437 installed rules files.
439 * udev will now create /dev/input/by-{id,path}/* style symlinks for
440 hidraw devices too. (Previously these would be created for other
441 input device types only.)
443 * *.link files gained support for configuring various Energy Efficient
444 Ethernet (EEE) settings in a new [EnergyEfficientEthernet] section.
446 * udevadm test gained a new --json= switch for generating JSON output.
448 * A new udev builtin "factory_reset" has been added that simply reports
449 if the system is currently booted in factory reset mode. This can be
450 used by udev rules that determine the location of the root file
451 system, in order to decide whether to expect that a root file already
452 exists or still needs to be created/formatted/encrypted.
454 * The "blkid" builtin of udev has been changed to determine the host
455 root file system by looking for the used ESP/XBOOTLDR only while
456 running in the initrd. When running after the initrd→host transition
457 it now just uses the root file system already mounted to /. Of
458 course, usually this should have the same results, but there are
459 situations thinkable where the ESP is on one disk and the root fs on
460 another, and we better not second guess this once we transitioned
461 onto the root file system.
463 * A new udev builtin "dissect_image" has been added that uses the usual
464 DDI image dissection code to identify partitions and their use and
465 relationships. This is used by new udev rules to generate a set of
466 symlinks in /dev/disk/by-designator/ that point to the various
467 discovered partitions by their designator.
469 * Android debug USB interfaces (ADB DbC, ADB, Fastboot) are now
470 automatically marked for unprivileged access, generically via a new
471 ID_DEBUG_APPLIANCE= udev property. Or in other words, running "adb"
472 again your Android phone connected via USB, set to debug mode should
473 just work without any additional rules.
475 * A new standard group "clock" has been introduced that is now used by
476 default for PTP and RTC device nodes in /dev/.
480 * The [Network] section in .network fails gained a new setting
481 MPLSRouting= which may be used to control Multi-Protocol Label
482 Switching on an interface.
484 * A system-wide default for ClientIdentifier= may now be set in
485 networkd.conf. (Previously this had to be configured individually in
488 * PersistLeases= setting in [DHCPServer] section now also accepts
489 "runtime", to make the DHCP server saves and loads bound leases on
492 * A new Preference= setting has been added to the [IPv6RoutePrefix]
493 section to configure the route preference field.
495 * New LinkLocalLearning=, Locked=, MACAuthenticationBypass=,
496 VLANTunnel= settings have been added the [Bridge] section of .network
499 * .netdev files gained new External=/VNIFilter= settings in [VXLAN]
502 * .netdev files can now configure HSR/SRP network devices too, via he new
505 * The LLDP client will now pick up the VLAN Id from LLDP data. The LLDP
506 sender will now send this field on VLAN devices.
508 * The DHCPv4 client in systemd-networkd now also supports BOOTP (via
509 the new BOOTP= setting).
511 sd-varlink & sd-json:
513 * An API call sd_varlink_reset_fds() has been added that undoes the
514 effect of sd_varlink_push_fd() (the API for submitting file
515 descriptors to send along with a method call), without actually
516 sending a Varlink message.
518 * An API call sd_varlink_server_listen_name() has been added that is
519 just like sd_varlink_server_listen_auto() but takes one additional
520 parameter: the file descriptor name (in the sense of $LISTEN_FDNAMES)
521 to look for, instead of "varlink". This is useful for services that
522 implement multiple Varlink services on distinct sockets and shall be
523 activatable through either.
525 * A pair of API calls sd_json_variant_type_from_string() and
526 sd_json_variant_type_to_string() have been added that may be used to
527 convert the JSON variant type identifier into a string representation
530 * A pair of API calls sd_varlink_get_input_fd() and
531 sd_varlink_get_output_fd() have been added that allow querying the
532 connection file descriptors individually for each direction, in case
533 two distinct file descriptors are used (for example in stdin/stdout
536 * A new API call sd_varlink_get_current_method() has been added which
537 reports the method call name currently being processed.
539 * Two new flags SD_VARLINK_SERVER_ALLOW_FD_PASSING_INPUT and
540 SD_VARLINK_SERVER_ALLOW_FD_PASSING_OUTPUT have been defined, which
541 may be passed to sd_varlink_server_new(), and ensure that any
542 connections associated with the server instance are automatically
543 created with file descriptor passing enabled for input or output.
545 * The "io.systemd.System" fallback Varlink errors that sd-varlink
546 generates for Linux 'errno' style error numbers now carry both the
547 numeric value (as before) and the symbolic name (i.e. "ENOENT"),
548 ensuring that the error remains somewhat portable (as the numeric
549 values are Linux and possibly architecture-specific).
551 * The generic "io.systemd.service" Varlink service that various of our
552 long-running services implement, gained a new GetEnvironment() call
553 that returns the current environment block of the service's main
554 process. In addition, this service interface has been implemented in
555 many more long-running services.
557 * A new sd-varlink call sd_varlink_get_description() has been added
558 that returns the string previously set via
559 sd_varlink_set_description().
561 * A new sd-varlink API call sd_varlink_get_n_fds() has been added that
562 returns the number of pending incoming file descriptors on the
565 * A new flag SD_VARLINK_SERVER_MODE_MKDIR_0755 may now be ORed into the
566 mode parameter of sd_varlink_server_listen_address(). If specified
567 then any leading directories in the provided AF_UNIX socket path are
568 automatically created with an 0755 access mode, should they be
571 * sd_varlink_idl_parse() and sd_varlink_interface_free() have been
572 added to sd-varlink, which can be used to parse Varlink IDL data.
576 * varlinkctl gained a new --exec switch. When used a command line of a
577 command to execute once a Varlink method call reply has been received
578 may be specified. The command will receive the method call reply on
579 standard input in JSON format, and any passed file descriptors via
580 the $LISTEN_FDS protocol. This is useful for invoking method calls
581 that return file descriptors from shell scripts.
583 * varlinkctl gained a new --push-fd= switch which may be used to issue
584 a Varlink method call and send along one or more file descriptors on
585 transports that support it (i.e. AF_UNIX).
589 * A new API call sd_device_enumerator_add_all_parents() has been added
590 that may be used to include all parent devices of otherwise matching
591 devices in the enumeration.
593 * A new API call sd_device_get_sysattr_value_with_size() has been added
594 that returns a sysfs attribute file in binary form along with its
599 * A new configuration knob WallMessages= has been added to logind.conf,
600 which may be used to control whether wall(1) style messages shall be
601 sent to all consoles when the system goes down.
603 * A new pseudo session class "none" has been defined. This may be used
604 with the class= parameter of pam_systemd.so (and some other places)
605 to disable allocation of a systemd-logind session for a specific
606 session. Note that this is not a recommended mode of operation, as
607 such "ghost" sessions will not be properly accounted for, and are
608 excluded from the per-user/per-session resource accounting.
610 * Two new session classes "user-light"/"user-early-light" have been
611 added, that are just like the regular "user"/"user-early" session
612 classes, but differ in one way: they do not cause activation of the
613 per-user service manager. These new session classes are now used for
614 logins of non-regular users which are used in a non-interactive way.
616 * The pidfd inode ID of a session's leader process is now exposed as
617 D-Bus property for session objects, in addition to the PID. The inode
618 ID is a 64bit unique identifier for a process that is not vulnerable
623 * When issuing parallel A and AAAA lookups for the same domain name,
624 and one succeeds quickly, we'll now shorten the timeout for the
625 other. This should improve behaviour with DNS servers whose IPv6
626 support is flaky and reply to A quickly but not at all to AAAA.
628 * The "Monitor" Varlink IPC API of systemd-resolved now gained support
629 for a new SubscribeDNSConfiguration() call that enables subscription
630 to any DNS configuration changes, as they happen.
632 * systemd-networkd-wait-online gained a new --dns switch that ensures
633 that not only network connectivity is available, but also DNS
634 configuration is established in systemd-resolved, making use of the
635 new, aforementioned Varlink interface.
637 * resolved.conf gained a new setting RefuseRecordTypes= which takes a
638 list of RR types for which to refuse lookup attempts. This may be
639 used to for example block A or AAAA lookups on IPv4- or IPv6-only
642 * A new DNS "delegate zone" concept has been introduced, which are
643 additional lookup scopes (on top of the existing per-interface and
644 the one global scope so far supported in resolved), which carry one
645 or more DNS server addresses and a DNS search/routing domain. It
646 allows routing requests to specific domains to specific servers.
647 Delegate zones can be configured via drop-ins below
648 /etc/systemd/dns-delegate.d/*.dns-delegate.
650 * "resolvectl query -t sshfp" will now decode the returned RR
651 information, and show the cryptographic algorithms by name instead of
656 * The system hardware's serial number may now be read from DeviceTree
657 too, in addition to the existing SMBIOS/DMI based logic.
659 * New properties for the Chassis Asset Tag, the hardware SKU, and the
660 hardware version are now provided (backed by SMBIOS/DMI).
662 * hostnamed also exposes properties now for the image ID and image
663 version (this is very useful on image-based systems).
665 systemd-stub, systemd-boot & bootctl:
667 * UEFI firmware images may now be embedded in UKIs (in an ".efifw" PE
668 section), for use in bring-your-own-firmware scenarios in
669 Confidential Computing. The firmware is matched via CHIDs to the
670 local invoking VM, in a fashion conceptually close to the DeviceTree
671 selection already available since v257. If a suitable firmware image
672 is found at boot, and the system's firmware version does not match it,
673 the update is applied and the system is rebooted. If the firmware
674 matches, boot proceeds as usual.
676 * When systemd-stub is invoked through a network boot provided UKI, it
677 will now query the source URL and write it to the LoaderDeviceURL EFI
678 variable. This may then be used by Linux userspace to look for
679 further resources (such as a root disk image) at the same location.
681 * systemd-boot now understands two new Boot Loader Specification Type #1
682 stanzas: "uki" and "uki-url", which is very similar to "efi" and
683 "linux", and references an UKI, the latter on a remote HTTP/HTTPS
684 server. The latter is particularly relevant for implementing a fully
685 UKI based boot process, but with network provided UKI images.
687 * systemd-boot now looks for the special SMBIOS Type #11 vendor strings
688 io.systemd.boot.entries-extra=, and synthesizes additional boot menu
689 entries from the provided data. This is useful with systemd-vmspawn's
690 --smbios11= switch, see below.
692 * systemd-stub now defaults to a minimum of 120 available PE sections,
693 instead of the previous default of 30. This reflects the fact that
694 multi-profile UKI typically require a lot more sections than
695 traditional single-profile UKIs. Note that this is just a
696 compile-time default, downstream distributions might choose to raise
697 this further – in particular on ARM systems where many Devicetree
698 blobs shall be embedded into an UKI.
700 * systemd-boot's loader.conf configuration file gained a new
701 "reboot-on-error" setting which controls what to do if booting a
702 selected entry fails, i.e. whether to reboot or just show the menu
705 * bootctl's --no-variables switch has been replaced by
706 --variables=yes/no. By setting --variables=yes modification of EFI
707 variables can be forced now in environments where we'd previously
708 automatically turn this off (e.g. in choot() contexts).
710 * systemd-stub gained support for a couple of "extension" CHIDs, that
711 are not part of the Microsoft's original spec, and which include EDID
712 display identification information in the hash. This may be used to
713 match Devicetree blobs in UKIs. "systemd-analyze chid" has been
714 updated to support these extension CHIDs, too. (They are clearly
715 marked as extensions CHIDs, to emphasize they are systemd's own
716 invention, and not based on the Windows CHID spec.)
718 * systemd-boot's loader.conf configuration file gained a new
719 secure-boot-enroll-action setting which controls the action to take
720 once automatic Secure Boot keys have been enrolled, i.e. whether to
721 reboot or whether to shut down the system.
723 * There's a new LoaderSysFail EFI environment variable that userspace
724 may set to an entry match pattern for systemd-boot. If set, and the
725 system firmware reports some kind of system failure (for now this is
726 pretty much only about failed firmware updates) the selected entry is
727 booted into, instead of following the usual entry selection
728 logic. bootctl gained a new "set-sysfail" verb to set this variable.
730 systemd-nsresourced & systemd-mountfsd:
732 * When a new user namespace is registered and a name for it must be
733 supplied, this name may now optionally be mangled automatically so
734 that it follows the naming rules for namespaces employed. This makes
735 it easier to provide suitable identifiers to the service, without any
736 client-side preparations or clean-ups, and thus ensures allocation of
737 a userns can ultimately "just work".
739 * A special, fixed UID/GID range has been defined called the "foreign"
740 UID/GID range. It's intended to be used to persistently own
741 bootable OS/container images on disk (i.e. OS trees that use a
742 UID/GID assignments not local to the host, but "foreign", i.e. they
743 have their own /etc/passwd + /etc/group table or similar database),
744 so that they can be mapped to other user namespace UID/GID ranges at
745 runtime through ID-mapped mounts.
747 * systemd-mountfsd gained a new IPC call accessible to unprivileged
748 clients for acquiring an ID-mapped mount for any OS/container
749 directory tree which is itself owned by the foreign UID/GID range,
750 and has a parent directory owned by the caller's UID. This means the
751 systemd-nsresourced/systemd-mountfsd combination is now suitable for
752 running unprivileged containers both from a disk image and from a
755 * When activating a DDI via mountfsd's MountImage() call the returned
756 data will now include the literal path to attach each returned path
757 to, to simplify implementation of clients.
759 * systemd-nsresourced gained an API for allocating a network TAP device
760 to associate with a user namespaces. This can be used by unprivileged
761 VMMs, to acquire IP networking. The network interface associated with
762 the TAP device comes with a matching .link and .network file, so that
763 systemd-networkd will set up IP routing (with masquerading) on it
766 * systemd-nsresourced will now always ask polkit for authorization of
767 its operations, even if they are supposed to be accessible to
768 unprivileged clients, so that the PK policy has the last word.
770 * systemd-nsresourced gained a new API call MakeDirectory(), which
771 creates a new directory, owned by the foreign UID range. It's
772 supposed to be used in conjunction with MountDirectory() for creating
773 and populating new container trees within user/$HOME context.
777 * Support for unprivileged invocation of container images stored in
778 plain directories has been added, using the new IPC APIs provided by
779 "systemd-mountfsd", see above.
781 * systemd-nspawn's --private-users= switch now supports a new value
782 "managed", which will ensure allocation of a userns via
783 systemd-nsresourced, even if run privileged.
785 * If systemd-nspawn is used interactively, two new special key
786 sequences can be used to trigger an immediate clean shutdown or
787 reboot of the container with systemd running as PID 1: '^]^]p' for
788 shutdown and '^]^]r' for reboot. This is in addition to the
789 previously supported '^]^]^]' which triggers immediate shutdown
790 without going through the usual shutdown logic.
792 * systemd-nspawn will now invoke the TTY password agent if invoked
793 interactively and without privileges. This makes sure unprivileged
794 containers start to work even when no other polkit agent is currently
795 running for the user. The usual --no-ask-password switch is now also
796 available in systemd-nspawn to disable this.
798 * systemd-nspawn gained a new --bind-user-shell= switch which allows to
799 tweak the shell field of users bound into a container with
804 * systemd-machined now provides a comprehensive Varlink IPC API to its
807 * The pidfd inode ID of a machine's leader process is now exposed as
808 D-Bus property for machine objects, in addition to the PID. The inode
809 ID is a 64bit unique identifier for a process that is not vulnerable
812 systemd-measure, ukify, systemd-keyutil, systemd-sbsign:
814 * systemd-measure gained a new "policy-digest" verb. It's a lot like
815 "sign" but instead of calculating the right TPM policy digest for a
816 specific UKI to sign and then signing it, it leaves the latter step
817 out. This is useful to implement offline signing of the policy digest
818 of UKIS. ukify gained a --policy-digest option that exposes this
821 * ukify gained a new --sign-profile= switch for signing a specific UKI
822 profile (to support multi-profile UKIs).
824 * ukify gained a pair of --join-pcrsig= and --pcrsig= options which is
825 useful for offline signing TPM PCR policies, as it allows inserting
826 pre-prepared PCR signature blobs into a UKI.
828 * ukify gained a new --pcr-certificate= switch that takes the path to
829 an X.509 certificate to use in place of a PEM public key, as provided
830 via the existing --pcr-public=.
832 * systemd-keyutil gained a new verb "pkcs7" which can be used to
833 convert between PKCS#1 and PKCS#7 signatures. The --content= switch
834 may be used to generate inline signatures (as opposed to the default
835 of detached signatures). It also gained a new --hash-algorithm=
836 switch to select the hash algorithm for signatures.
838 * systemd-sbsign learnt support for offline SecureBoot signing via
839 --prepare-offline-signing, --signed-data=, --signed-data-signature=.
843 * A new PCR phase string is now measured into PCR 11 when storage
844 target mode is entered, ensuring that access to TPM key material can
845 be taken away, once storage target mode is activated.
847 * Similarly, a new string is measured when booting into factory reset
850 * A new service systemd-tpm2-clear.service has been introduced that can
851 be used to request clearing of the local TPM on next reboot. It comes
852 with a kernel command line option systemd.tpm2_allow_clear= that
853 controls its effect. The unit is hooked into the generic
854 factory-reset.target unit, so that it can do its thing when a factory
857 * If systemd-pcrextend (i.e. the tool making the various userspace TPM
858 PCR measurements) fails to do its thing, an immediate reboot is now
859 triggered, ensuring that somehow making PCR extensions fails cannot
860 be used to gain access to TPM objects to which access should have
861 been blocked already via PCR measurements.
863 * systemd-pcrlock gained a new "is-supported" verb that determines
864 whether local TPM and system provide all necessary functionality for
865 systemd-pcrlock to work. It does a superset of the checks
866 "systemd-analyze has-tpm2" does, and additionally ensures that the
867 TPM supports PolicyAuthorizeNV and SHA-256.
869 systemd-userdbd & systemd-homed:
871 * User records now support a new field "aliases" that may list
872 additional names the user record shall be accessible under. Any
873 string listed in the "aliases" array may be used wherever and
874 whenever the primary name may be used too, for example when logging
875 in. systemd-homed and in particular homectl have been updated to
876 support configuration of such alias names.
878 * If a user record has an initialized "realm" field, then the record
879 may now be referenced via the primary user name or any alias name,
880 suffixed with "@" and the realm, too.
882 * User records gained new fields tmpLimit, tmpLimitScale, devShmLimit,
883 devShmLimitScale which enforce quota on /tmp/ and /dev/shm/ at login
884 time, either in absolute or in relative values. These values default
885 to 80% for regular users, ensuring that a single user cannot easily
886 DoS a local system by taking away all disk space in /tmp/. The
887 homectl tool has been updated to make these new fields configurable.
889 * The userdb Varlink interface has been extended to support server-side
890 filtering by UID/GID min/max, fuzzy name matching and user
891 disposition. Previously this was supported by the userdbctl
892 client-side only. With this, userdb providers may now optionally
893 implement this server-side too in order to optimize the lookups.
895 * User records now support a concept of home "areas",
896 i.e. subdirectories of the primary $HOME directory that a user can
897 log into. This is useful to maintain separate development
898 environments or configuration contexts, but within the ownership of
899 the same user. Support for this is implemented in systemd-homed, but
900 is conceptually open to other backends, too.
902 New home areas can be created via "mkdir -p ~/Areas/ && cp /etc/skel
903 ~/Areas/foo", or removed by "rm -rf ~/Areas/foo". Whenever prompted
904 for login and a user name is requested, it is possible to enter a
905 username suffixed by "%" and the area name in order to log into the
906 specified area of the user. (e.g. "bar%foo"). Effectively this
907 ensures that $HOME and $XDG_RUNTIME_DIR include the area choice after
908 login. Note that at this moment it's not possible to log into a full
909 graphical session with this, since we'd have to start a per-area user
910 service manager for that, and we currently do not do this. But we
911 hope to provide this in one of the next releases. In order to
912 implement all this user records gained a new "defaultArea" field,
913 which is configurable with homectl's --default-area= switch.
915 * An explicit MIME type application/x.systemd-home is now used for all
916 LUKS *.home files managed by systemd.
918 * userdbctl gained a new switch --from-file=. If used the tool will not
919 look up a user or group record from the system's user database but
920 instead read it from the specified JSON file, and then present it in
921 the usual, human-readable fashion.
923 * systemd-homed gained D-Bus API calls for listing, adding, removing and
924 showing use record signing keys.
926 * homectl gained the verbs "list-signing-keys", "get-signing-key",
927 "add-signing-key", "remove-signing-key" and a switch
928 --key-name=. These may be used to easily make a single home directory
929 usable on multiple systems. A system credential
930 home.add-signing-key.* has been added that allows provisioning such
931 user record signing keys at boot.
933 * homectl gained a new switch "--dry-run" which can be used when
934 registering/creating users, and which will show the user record data
935 before it's submitted to systemd-homed. The tool will then terminate
936 before the submission.
938 * User/group records' perMachine section now support negative matches
939 too (i.e. for settings that apply to all systems but some selected
942 * systemd-homed gained a bus API call AdoptHome() for "adopting" a
943 .home file or .homedir directory from a foreign system
944 locally. homectl added a verb "adopt" exposing the new call. Together
945 with the signing key management functionality described above it
946 makes it very easy to migrate homes between systems.
948 * systemd-homed gained two new bus API calls RegisterHome() and
949 UnregisterHome() for registering a home locally by providing just the
950 user record, without any logic to actually create the home
951 directory. homectl gained "register" and "unregister" verbs exposing
952 this. This is useful for registering network user accounts locally,
953 i.e. where some foreign user record and home directory already exists
954 on some server, and just need to be registered locally. This can be
955 used like the following to make a local systemd-homed home directory
956 securely accessible from some other system:
958 homectl update lennart --ssh-authorized-keys=… -N --storage=cifs --cifs-service="//$HOSTNAME/lennart"
959 homectl get-signing-key | ssh targetsystem homectl add-signing-key --key-name="$HOSTNAME".public
960 homectl inspect -E lennart | ssh targetsystem homectl register -
961 ssh lennart@targetsystem
963 There's also a system credential home.register.* now that can execute
964 the registration operation for a provided user record automatically
967 * homectl gained a new switch --seize= taking a boolean argument. If
968 true when used together with the "create" or "register" verbs any
969 cryptographic signature information is stripped from the user record,
970 taking over the user record for local ownership. This switch is
971 useful when migrating a home directory to a different host, without
972 retaining the relationship to the originating host.
974 * homectl gained a new --match= switch which allows to generate
975 accounts with perMachine matching sections.
977 * userdbctl gained a new verb "load-credentials", with a service unit
978 systemd-userdb-load-credentials.service which invokes it. When
979 invoked this command will look for any passed credentials named
980 userdb.user.* or userdb.group.*. These credentials may contain
981 user/group records in JSON format. They will be copied into
982 /run/userdb/ (where static userdb JSON records can be placed), with
983 the appropriate symlink from the UID/GID added in, as any membership
984 relationships between user/groups replicated as .membership files. Or
985 in other words: it's very easy to provision a complete user/group
986 record in an invoked system, by providing the user/group JSON record
987 as system credential. Note that these credentials are unrelated to
988 similar credentials supported by systemd-homed. "userdb
989 load-credentials" creates "static" user records via drop-in files in
990 /run/userdb/ (and thus covers system users and suchlike) while
991 systemd-homed creates only systemd-homed managed use (i.e. only
994 * User/group records gained a new "uuid" field that may be used to
995 place an identifying UUID in the record.
997 systemd-run and run0:
999 * run0 gained a new --lightweight= switch which controls whether to
1000 pull in a service manager for the target session (i.e. this
1001 ultimately chooses between the "user"/"user-early" session class on
1002 one hand or the "user-light"/"user-early-light" session class on the
1005 * systemd-run gained a new --job-mode= switch for controlling the job
1006 mode when enqueuing the start job for the transient unit. This is
1007 similar to the switch of the same name of "systemctl start".
1009 * run0 gained a new --area= switch for directly entering a specific
1010 home area (see above).
1012 * systemd-run/run0 gained a new --pty-late switch that is just like
1013 --pty but sets up TTY forwarding only once the unit is fully
1014 activated. This is relevant for avoiding TTY ownership collisions between
1015 the TTY forwarding and potential password queries using the
1016 systemd-ask-password infrastructure. run0 now defaults to this mode for
1017 interactive operations.
1019 * The --chdir= switch now accepts the special value '~' to force
1020 changing into the target user's home directory.
1022 * run0 gained a new --via-shell switch that ensures any specified
1023 command is invoked via the target user's shell instead of directly.
1025 DDI support & systemd-dissect:
1027 * systemd-dissect gained a new --loop-ref-auto switch which initializes
1028 the --look-ref= field from a suitable string derived from the DDI
1031 * systemd-dissect's --attach command now supports a new --quiet switch
1032 that suppressed output of the loopback device node path that is
1035 * A generic service template systemd-loop@.service has been added that
1036 wraps "systemd-dissect --attach", and attaches a disk image whose
1037 path is encoded in the instance identifier of the unit to a new
1038 loopback block device. This may be used to attach arbitrary disk
1039 images to loopback devices at boot.
1041 * There's now a per-user counterpart of /var/lib/machines/ defined as
1042 ~/.local/state/machines/. Various tools such as systemd-nspawn +
1043 systemd-vmspawn now will search this directory when looking for a
1044 disk image, when invoked in unprivileged user context.
1045 systemd-dissect's --discover command may now be combined with --user
1046 or --system to choose in which of the directory scopes to look for
1049 * systemd-dissect gained a new --all switch. If specified the tool will
1050 not just discover DDIs (i.e. disk images) but also images stored in
1051 regular directories.
1053 * systemd-dissect gained a new "--shift" switch for recursively
1054 re-chown()ing a directory tree from one set of UID/GIDs to another.
1055 This may be used to shift a tree from the base-0-UID range to the
1056 foreign UID range or back.
1058 * systemd-dissect gained new --usr-hash= and --usr-hash-sig= options,
1059 that are similar to the existing --root-hash=/--root-hash-sig=
1060 options, but for the /usr/ partition. This allows the root hash of
1061 the /usr/ Verity volume and its signature to be specified.
1063 * When dissecting/mounting a DDI disk image, and no Verity root hash or
1064 signature is provided, suitable values are now automatically
1065 discovered from the image itself.
1067 * systemd-gpt-auto-generator now understands root=dissect and
1068 mount.usr=dissect as kernel command line options that explicitly
1069 request the full blown DDI dissector to be used to discover the root
1070 and /usr/ file system, including automatic Verity root hash and
1071 signature discovery, automatic handling of versioning, image policy
1072 enforcement and filtering and so on.
1074 * The DDI dissection logic now understands a concept of partition
1075 "filtering". A partition filter is simply a per-designator globbing
1076 pattern to match the partition labels against. This may be used
1077 support parallel installations of multiple operating systems on the
1078 same disk, where each OS names its partitions with a specific prefix
1079 or similar. systemd-dissect gained a new --image-filter= switch to
1080 configure this filter. The new "dissect_image" udev plugin and
1081 systemd-gpt-auto-generator now understand the new
1082 systemd.image_filter= kernel command line switch configuring this
1083 filter for the system.
1085 systemd-importd & importctl:
1087 * systemd-pull/importctl now supports ASCII armored (*.asc) GPG signatures.
1089 * The systemd.pull= and rd.systemd.pull= kernel command line switches
1090 (which may be used to automatically download a VM, container, confext,
1091 or sysext at boot) now understand a new flag "blockdev". When
1092 specified the downloaded image is attached to a loopback block device
1093 after download. This may be used to boot directly into a disk image
1094 downloaded via HTTP via a kernel command line like this:
1096 rd.systemd.pull=raw,machine,verify=no,blockdev:image:https://192.168.100.1:8081/image.raw root=/dev/disk/by-loop-ref/image.raw-part2
1098 * systemd.pull=/rd.systemd.pull= also gained support for a new flag
1099 "bootorigin". If specified and if the system was network booted
1100 through systemd-stub (which now sets the LoaderDeviceURL EFI
1101 variable, see above), the URL to boot from is now automatically
1102 formed from the UKI network boot URL with a new suffix. Example:
1104 rd.systemd.pull=raw,machine,verify=no,blockdev,bootorigin:rootdisk:image.raw.xz root=/dev/disk/by-loop-ref/rootdisk.raw-part2
1106 * The systemd.pull=/rd.systemd.pull= switches now also support a new
1107 flag "runtime=", taking a boolean argument. If true the downloaded
1108 image is placed below the /run/ hierarchy instead of /var/. It
1109 defaults to true for rd.systemd.pull= (i.e. for downloads made in the
1110 initrd), and false for systemd.pull= (i.e. for those made after the
1111 initrd→host transition).
1113 * New generic target units imports-pre.target and imports.target have
1114 been introduced that are ordered before and after all downloads.
1116 * systemd-importd gained support for downloading images compressed with
1117 zstd now, too. (In addition to .xz, .gz and .bz2.)
1121 * A new --smbios11= switch may be used to pass an SMBIOS Type #11
1122 vendor string easily into the booted process. This has various uses,
1123 one of them is to add additional menu entries to systemd-boot for a
1124 specific invocation. Example:
1126 --smbios11=io.systemd.boot.entries-extra:particleos-current.conf=$'title ParticleOS Current\nuki-url http://example.com/somedir/uki.efi'
1128 * A new switch --grow-image= has been added taking a size in bytes. If
1129 specified the image booted into is grown to the specified size if
1130 found to be smaller.
1132 * systemd-vmspawn supports unprivileged network now, using
1133 systemd-nsresourced's new API to acquire a TAP network device
1136 * A new --tpm-state= setting allows precise control of TPM state
1141 * A new tool systemd-factory-reset has been added that may be used to
1142 request or cancel a factory reset request for the next reboot. It is
1143 also accessible via its own Varlink API.
1145 * A new target unit factory-reset-now.target has been added that
1146 executes an immediate factory reset. (Previously factory-reset.target
1147 existed already that requested it for next reboot).
1149 * A new kernel command line option systemd.factory_reset= has been
1150 added for explicitly requesting a factory reset. (Implemented via a
1151 new systemd-factory-reset-generator)
1153 * A new document explaining the factory reset logic in detail has been
1154 added. It is available online here:
1156 https://systemd.io/FACTORY_RESET
1160 * systemd-repart gained a new switch --join-signature= for supporting
1161 offline Verity signing.
1163 * systemd-repart gained a new switch --append-fstab= for controlling
1164 how to write or append automatically generated /etc/fstab entries.
1166 * CopyFiles= lines can now contain an "fsverity=copy" flag to preserve
1167 the fs-verity status of the source files when populating the
1170 * systemd-repart has been updated to automatically generate the
1171 extended attributes systemd-validatefs@.service understands (see
1172 below), for all partitions it recognizes. Controllable via the
1173 AddValidateFS= partition setting (which defaults to true).
1175 * repart.d/ drop-ins gained a new setting FileSystemSectorSize= which
1176 allows configuring the sector size that file systems for newly
1177 formatted file systems explicitly.
1179 * systemd-repart will now enforce a minimum size for ESP/XBOOTLDR
1180 partitions of 100M (on 512b sector drives) or 260M (on 4K sector
1181 drives), in accordance to the requirements for these kind of
1184 * The Format= setting in repart.d/ files gained support for a special
1185 value "empty". This is a shortcut to set up an empty partition and
1186 set the partition label to "_empty", and set the "NoAuto" GPT
1187 flag. The former is useful as systemd-sysupdate recognizes empty
1188 partitions that way, the latter is useful to ensure that the
1189 partition is not automatically made used of as is, on any OS that
1194 * systemd-ask-password now provides a small Varlink API to
1195 interactively query the user for a password using the usual agent
1196 logic. This makes it easier for external programs (for example
1197 daemons) to query for boot-time passwords and similar, using
1198 systemd's infrastructure.
1200 * The logging logic in systemd's codebase now implements the
1201 DEBUG_INVOCATION= interface added to service management in v257. Or
1202 in other words: the RestartMode=debug setting may now be added for
1203 any of systemd's own service and has the intended effect of enabling
1204 debug logging if it gets automatically restarted.
1206 * systemd-analyze gained a new "chid" verb, which shows the "Computer
1207 Hardware IDs" (CHIDs) of the local system. This is useful for
1208 preparing CHID-to-DeviceTree mappings when building UKIs.
1210 * systemd-analyze gained a new "transient-settings" verb, which shows
1211 all unit settings one can configure dynamically via the "-p" switch
1212 when invoking transient units.
1214 * systemd-analyze gained a new "unit-shell" verb that invokes an
1215 interactive shell inside the namespaces of the main process
1216 of a specified unit. This is useful for debugging unit sandboxes, and
1217 getting an idea how things look like from the "inside" of a service.
1219 * The "package note" specification ELF binaries has been extended to
1220 cover PE binaries (i.e. UEFI binaries), too.
1222 * New kernel command line parameters systemd.break= and
1223 rd.systemd.break= have been introduced that insert interactive (as
1224 in: shell prompt) "breakpoints" into the boot process at various
1225 locations, in order to simplify debugging. For now four breakpoints
1226 are defined: "pre-udev", "pre-basic", "pre-mount",
1227 "pre-switch-root". Similar functionality has previously existed in
1228 the Dracut initrd generator, but is generalized with this new
1229 concept, and extended to the post-switch-root boot phases.
1231 * The systemd-path tool now learnt new paths for the per-system and
1232 per-user credential store.
1234 * A new tool systemd-pty-forward has been added that allocates a pseudo
1235 TTY ("PTY") and invokes a process on it, forwarding any output to the
1236 TTY it is invoked on. It can optionally apply background coloring and
1237 suchlike, and is mostly just a separate tool that makes the PTY
1238 forwarding logic used in systemd-nspawn, systemd-vmspawn, run0
1239 available separately.
1241 * systemd-oomd can now reload its configuration at runtime, following
1242 the usual protocols.
1244 * systemd-detect & ConditionVirtualization= now recognize the "Arm
1245 Confidential Compute Architecture" (cca) confidential virtualization.
1247 * systemd-sysusers now generates Linux audit records when it adds
1250 * systemd-firstboot's interactive prompts for locale or keymaps now
1251 support tab completion.
1253 * systemd-mount gained support for a new --canonicalize= switch that
1254 may be used to turn off client-side path canonicalization before
1255 trying to unmount some path.
1257 * systemd-notify gained a new --fork switch which inverts the role that
1258 systemd-notify plays in the sd_notify() protocol: instead of sending
1259 out notification messages, it will listen for them, forking off a
1260 command that is expected to send them. Once READY=1 is received
1261 systemd-notify will exit, leaving the child running. This is useful
1262 for correctly forking off processes that implement the sd_notify()
1263 protocol from shell scripts.
1265 * systemd-fstab-generator now supports a root=bind:… syntax for
1266 creating bind mounts for the root file system. This is useful for
1267 booting into tarballs downloaded at boot. As an example, consider a
1268 kernel command line like this:
1270 rd.systemd.pull=tar,machine,verify=no:root:http://192.168.100.1:8081/image.tar root=bind:/run/machines/root ip=any
1272 * libapparmor is now loaded via dlopen() instead of using direct shared
1273 library linking. This allows downstream distributions to provide AA
1274 support as a runtime option instead of making the AA userspace a
1275 mandatory dependency.
1277 * A new generic remote-integritysetup.target unit has been added that
1278 matches remote-veritysetup.target and remote-cryptsetup.target's role
1279 for remote block devices, but for dm-integrity devices.
1281 * A new document about finding boot components and the root disk of the
1282 OS has been added. It's available online here:
1284 https://systemd.io/ROOTFS_DISCOVERY
1286 * Whenever any systemd tool begins or ends a new TTY context (i.e. takes
1287 over a TTY for some time) a new OSC sequence is now emitted, with
1288 various details about the context. This new OSC sequence can be
1289 interpreted by terminal emulators to visualize the context/source TTY
1290 output originates from or to show various kinds of metadata for
1291 it. The OSC sequence is specified in this document:
1293 https://systemd.io/OSC_CONTEXT
1295 Contexts are generated for systemd-nspawn/systemd-vmspawn boots, for
1296 run0 or systemd-run sessions, whenever PAM TTY sessions start or end,
1297 and when shell command executions start and end. Metadata sent along
1298 contains hostname, machine ID, boot ID, exit status, unit information
1301 * If PID 1 makes up a suitable $TERM for a TTY it activates a service
1302 on (in case there are no other hints on how to choose it) it will now
1303 also set $COLORTERM=truecolor. Moreover, if $COLORTERM or $NO_COLOR
1304 are set on the kernel cmdline we'll now import them into PID1's
1305 environment block, just like $TERM itself. Moreover, systemd-nspawn
1306 and run0 will now propagate $COLORTERM and $NO_COLOR from the calling
1307 to the target environment, if set, just like $TERM is already
1308 handled. Or to say this with different words: the triplet of $TERM,
1309 $COLORTERM, $NO_COLOR is now processed jointly and in similar ways,
1310 wherever appropriate.
1312 * systemd-update-done gained a new --root= switch to operate in
1313 "offline" mode on a specific file system tree.
1315 * A new template service systemd-validatefs@.service has been added
1316 that can validate usage of file systems. Specifically, it will look
1317 for certain extended attributes stored on the top-level directory
1318 inode of the mount, which may encode various constraints on use of
1319 the file system. For example, it may encode a directory path the file
1320 system must be mounted to, a GPT type UUID that must be used for the
1321 partition the file system is located in and more. This provides
1322 protection in case GPT auto-discovery is used to discover the mounts,
1323 but essential metadata outside of the file system itself has been
1324 tampered with. This operates under the assumption that the extended
1325 attributes on the root inode of the file system are protected by
1326 dm-verity or dm-crypt/dm-integrity, even if the GPT metadata has no
1327 equivalent cryptographic protection. If a file system carries these
1328 extended attributes but they do not match the current use and
1329 location of the file system an immediate reboot is triggered.
1331 * systemd-gpt-auto-generator now understands a new mount option
1332 x-systemd.validatefs for /etc/fstab entries. If specified an instance
1333 of systemd-validatefs@.service is automatically pulled in by the
1336 * systemd-fstab-auto-generator and systemd-gpt-auto-generator now
1337 understand root=off on the kernel command line which may be used to
1338 turn off any automatic or non-automatic mounting of the root file
1339 system. This is useful in scenarios where a boot process shall never
1340 transition from initrd context into host context.
1342 * systemd-ssh-proxy now supports an alternative syntax for connecting
1343 to SSH-over-AF_VSOCK, in order to support scp and rsync better: "scp
1344 foo.txt vsock%4711:" should work now. (The pre-existing syntax used
1345 "/" instead of "%" as separator, which is ambiguous in scp/rsync
1346 context even if not for ssh itself.)
1348 * "systemctl start" and related verbs now support a new --verbose
1349 mode. If specified the live log output of the units operated on is
1350 shown as long as the operation lasts.
1352 * sd-bus: a new API call sd_bus_message_dump_json() returns a JSON
1353 representation of a D-Bus message.
1355 * sd-daemon: a new call sd_pidfd_get_inode_id() has been added
1356 for acquiring the unique inode ID of a pidfd, coupling the
1357 $MAINPIDFDID/$MANAGERPIDFDID and session/machine leader pidfd IDs
1358 exposed as described above.
1360 * systemd-coredump will now attach a new COREDUMP_DUMPABLE= journal
1361 field to all coredumps indicating the "dumpable" per-process flag (as
1362 settable via PR_SET_DUMPABLE) at the moment the coredump took
1363 place. It will also add a new journal field COREDUMP_BY_PIDFD= that
1364 indicates whether the coredump was acquired via a stable pidfd to the
1367 * systemd-sysext (and portable services with sysexts applied) will now
1368 take the os-release "ID_LIKE=" field into account when validating that
1369 a sysext images is compatible with the underlying image. Previously
1370 it would only check "ID=".
1372 * A new UID range has been defined for "greeters", i.e. graphical login
1373 prompt UIs that shall be security isolated from each other. This is
1374 supposed to be used by graphical display managers (specifically:
1375 gdm), to ensure that it is harder to exploit the UI sessions used to
1376 prompt the user for login credentials, in order to gain access to the
1377 prompts of other users.
1379 * systemd-socket-activate gained a new --now switch which ensures the
1380 specified binary is immediately invoked, and not delayed until a
1381 connection comes in.
1383 * systemd-ssh-generator will now generate the AF_VSOCK ssh listener
1384 .socket unit, so that a tiny new helper "systemd-ssh-issue" is
1385 invoked when the socket is bound, that generates a drop-in file
1386 /run/issue.d/50-ssh-vsock.issue that is shown by "login" and other
1387 subsystems at login time. The file reports the AF_VSOCK CID of the
1388 system, along with very brief information how to connect to the
1389 system via ssh-over-AF_VSOCK. Or in other words: if the system is
1390 booted up in an AF_VSOCK capable VM the console login screen shown
1391 once boot-up is complete will tell you how to connect to the system
1392 via SSH, if that's available.
1398 Incompatible changes:
1400 * The --purge switch of systemd-tmpfiles (which was added in v256) has
1401 been reworked: it will now only apply to tmpfiles.d/ lines marked
1402 with the new "$" flag. This is an incompatible change, and means any
1403 tmpfiles.d/ files which shall be used together with --purge need to
1404 be updated accordingly. This change has been made to make it harder
1405 to accidentally delete too many files when using --purge incorrectly.
1407 * The systemd-creds 'cat' verb now expects base64-encoded encrypted
1408 credentials as input, for consistency with the 'decrypt' verb and the
1409 LoadCredentialEncrypted= service setting. Previously it could only
1410 read raw, unencoded binary data.
1412 * Support for automatic flushing of the nscd user/group database caches
1415 * The FileDescriptorName= setting for socket units is now honored by
1416 Accept=yes sockets too, where it was previously silently ignored and
1417 "connection" was used unconditionally.
1419 * systemd-logind now always obeys block inhibitor locks, where previously
1420 it ignored locks taken by the caller or when the caller was root. A
1421 privileged caller can always close the other sessions, remove the
1422 inhibitor locks, or use --force or --check-inhibitors=no to ignore the
1423 inhibitors. This change thus doesn't affect security, since everything
1424 that was possible before at a given privilege level is still possible,
1425 but it should make the inhibitor logic easier to use and understand,
1426 and also help avoiding accidental reboots and shutdowns. New 'block-weak'
1427 inhibitor modes were added, if taken they will make the inhibitor lock
1428 work as in the previous versions. Inhibitor locks can also be taken by
1429 remote users (subject to polkit policy).
1431 * systemd-nspawn will now mount the unified cgroup hierarchy into a
1432 container if no systemd installation is found in a container's root
1433 filesystem. $SYSTEMD_NSPAWN_UNIFIED_HIERARCHY=0 can be used to override
1436 * /dev/disk/by-id/nvme-* block device symlinks without an NVMe
1437 namespace identifier are now fixed to namespace 1 of the device. If
1438 no namespace 1 exists for a device no such symlink is
1439 created. Previously, these symlinks would point to an unspecified
1440 namespace, and thus not be strictly stable references to
1441 multi-namespace NVMe devices. These un-namespaced symlinks are mostly
1442 obsolete, users and applications should always use the ones with
1443 encoded namespace information instead. This change should not affect
1444 too many systems, because most NVMe devices only know a namespace 1
1447 * Support for cgroup v1 ('legacy' and 'hybrid' hierarchies) is now
1448 considered obsolete and systemd by default will ignore configuration
1449 that enables them. To forcibly reenable cgroup v1 support,
1450 SYSTEMD_CGROUP_ENABLE_LEGACY_FORCE=1 must additionally be set on the
1451 kernel command line.
1453 Announcements of Future Feature Removals:
1455 * The D-Bus method org.freedesktop.systemd1.StartAuxiliaryScope() is
1456 deprecated because accounting data and such cannot be reasonably
1457 migrated between cgroups. It is likely to be fully removed in a
1458 future release (reach out if you have use cases).
1460 * The recommended kernel baseline version has been bumped to v5.4
1461 (released in 2019). Expect limited testing on older kernel versions,
1462 where "old-kernel" taint flag would also be set. Support for them
1463 will be phased out in a future release in 2025, i.e. we expect to bump
1464 the minimum baseline to v5.4 then too.
1466 * The complete removal of support for cgroup v1 ('legacy' and 'hybrid'
1467 hierarchies) is scheduled for v258.
1469 * Support for System V service scripts is deprecated and will be
1470 removed in v258. Please make sure to update your software
1471 *now* to include a native systemd unit file instead of a legacy
1472 System V script to retain compatibility with future systemd releases.
1474 * To work around limitations of X11's keyboard handling systemd's
1475 keyboard mapping hardware database (hwdb.d/60-keyboard.hwdb) so far
1476 mapped the microphone mute and touchpad on/off/toggle keys to the
1477 function keys F20, F21, F22, F23 instead of their correct key codes.
1478 This key code mangling will be removed in the next systemd release.
1479 To maintain compatibility with X11 applications that rely on the old
1480 function key code mappings, this mangling has now been moved to the
1481 relevant X11 keyboard driver modules. In order to ensure these keys
1482 continue to work, update to xf86-input-evdev >= 2.11.0 and
1483 xf86-input-libinput >= 1.5.0 before updating to systemd >= 258.
1485 * Support for the SystemdOptions EFI variable is deprecated.
1486 'bootctl systemd-efi-options' will emit a warning when used. It seems
1487 that this feature is little-used and it is better to use alternative
1488 approaches like credentials and confexts. The plan is to drop support
1489 altogether at a later point, but this might be revisited based on
1492 * systemd-run's switch --expand-environment= which currently is disabled
1493 by default when combined with --scope, will be changed in a future
1494 release to be enabled by default.
1498 * systemd's JSON API is now available as public interface of
1499 libsystemd, under the name "sd-json". The purpose of the library is
1500 to allow structures to be conveniently created in C code and
1501 serialized to JSON, and for JSON to be conveniently deserialized into
1502 in-memory structures, using callbacks to handle specific
1503 keys. Various data types like integers, floats, booleans, strings,
1504 UUIDs, base64-encoded and hex-encoded binary data, and arrays are
1505 supported natively. The library has been part of systemd for a while
1506 as internal component, and is now made publicly available. One major
1507 user of sd-json is sd-varlink (see below). Note that the
1508 documentation of sd-json is very much incomplete for now, but the
1509 systemd codebase provides plenty real-life code examples.
1511 * systemd's Varlink IPC API is now available as part of libsystemd,
1512 under the name "sd-varlink". This library is a C implementation of
1513 the Varlink IPC system (https://varlink.org/) that has been adopted
1514 by systemd for various interfaces. It relies on the sd-json JSON
1515 component, see above. Note that the documentation of sd-varlink is
1516 very much incomplete for now, but the systemd codebase provides
1517 plenty real-life code examples.
1519 * sd-bus gained a new call sd_bus_pending_method_calls() which returns
1520 the number of currently open asynchronous method calls initiated on
1521 this connection towards peers.
1523 * sd-device gained a new call sd_device_monitor_is_running() that
1524 returns whether the specified monitor object is already running. It
1525 also gained sd_device_monitor_get_fd(),
1526 sd_device_monitor_get_events(), sd_device_monitor_get_timeout() and
1527 sd_device_monitor_receive() to permit sd-device to run on top of a
1528 foreign event loop implementation. It also gained
1529 sd_device_get_driver_subsystem() which returns the subsystem of
1530 driver objects. The new sd_device_get_device_id() call returns a
1531 short string identifying the device record.
1533 System and Service Management:
1535 * The environment variable $REMOTE_ADDR is now set when using
1536 per-connection socket activation for AF_UNIX stream sockets. It
1537 contains the AF_UNIX peer address of the connection. (Previously the
1538 environment variable was only set for IP sockets.)
1540 * Multipath TCP (MPTCP) is now supported as a socket protocol for
1543 * A new /etc/fstab option x-systemd.wants= creates "Wants="
1544 dependencies. (This is similar to the previously available
1545 x-systemd.requires=.)
1547 * The initialization of the system clock during boot and updates has
1548 been simplified: both PID 1 or systemd-timesyncd will pick the latest
1549 minimum time as indicated by the compiled-in epoch,
1550 /usr/lib/clock-epoch, and /var/lib/systemd/timesync/clock. See
1551 systemd(1) for an detailed updated description.
1553 * The kernel's Ctrl-Alt-Delete handling is re-enabled during late
1554 shutdown, so that the user may use it to initiate a reboot if the
1555 system freezes otherwise.
1557 * The new value "identity" for the unit setting PrivateUsers= may be
1558 used to request a user namespace with an identity mapping for the
1559 first 65536 UIDs/GIDs. This is analogous to the systemd-nspawn's
1560 --private-users=identity.
1562 * The new value "disconnected" for the unit setting PrivateTmp= may be
1563 used to specify that a separate tmpfs instance should be used for
1564 /tmp/ and /var/tmp/ for the unit.
1566 * The server manager (and various other tools too) use pidfds in more
1567 places to refer to processes.
1569 * A build option -D link-executor-shared=false can be used to build
1570 the systemd-executor binary (added in a previous release) in a way
1571 where it does not link to shared libsystemd-shared-….so library.
1572 PID1 holds a reference to the executor binary that was on disk when
1573 the manager was started or restarted, but the shared libraries it is
1574 linked to are not loaded until the executor binary needs to be used.
1575 This partial static linking is a workaround for the issue where,
1576 during upgrades, the old libsystemd-shared-….so may have already
1577 been removed and the pinned executor binary will just fail to
1580 * The systemd.machine_id= kernel command line parameter interpreted by
1581 PID 1 now supports an additional special value: if set to "firmware"
1582 the machine ID is initialized from the SMBIOS/DeviceTree system
1583 UUID. (Previously this was already done automatically in VM
1584 environments, this extends the concept to any system, but only on
1585 explicit request via this option.)
1587 * The ImportCredential= setting in service unit files now permits
1588 renaming of credentials as they are imported.
1590 * The RestartMode= setting gained a new "debug" value. If specified and
1591 the service fails so that it shall be restarted it is invoked in
1592 "debugging mode". Debugging mode means that the $DEBUG_INVOCATION
1593 environment variable will be set to "1" for the new
1594 invocation. Moreover, any setting LogLevelMax= will be temporarily
1595 changed to "debug" for the next invocation. This mode is useful to
1596 automatically repeat invocation of tools in case they fail – but with
1597 additional logging or testing routines enabled.
1599 * A new service setting BindLogSockets= has been added that
1600 controls whether the AF_UNIX sockets required for logging shall be
1601 bind mounted to the mount sandbox allocated for the service.
1603 * At early boot, PID 1 will now optionally load a policy for the new
1606 * Transient services (as invoked by the StartTransientUnit() D-Bus
1607 method) may now receive additional, arbitrary file descriptors to
1608 pass to executed service processes during activation using the new
1609 ExtraFileDescriptor= unit property.
1611 * Calendar .timer units gained a new boolean DeferReactivation=
1612 option. If enabled and the repetitive calendar timer elapses again
1613 while the service the timer activates is still running, immediate
1614 reactivation of the service once it finishes is skipped, and the
1615 timer has to elapse again before the service is reactivated.
1617 * Generator processes invoked by the service manager will now receive a
1618 new environment variable $SYSTEMD_SOFT_REBOOTS_COUNT that indicates
1619 how many times the system has been soft-rebooted since the kernel
1622 * A new service property ManagedOOMMemoryPressureDurationSec= has been
1623 added that complements the existing
1624 ManagedOOMMemoryPressureDurationLimit= and specifies the PSI
1625 measurement interval for the specific unit.
1627 * The sd_notify() protocol has been extended to allow changing the main
1628 PID of a process by providing a pidfd of the new main process, or by
1629 specifying the pidfd inode number. Previously this was only supported
1630 by specifying the classic UNIX PID, which of course is racy.
1632 * The SocketUser=/SocketGroup= settings of .socket units are now also
1633 applied to POSIX message queues.
1635 * The ProtectControlGroups= unit file setting now supports two
1636 additional values: if set to "private" a new cgroup namespace is
1637 allocated for the service and cgroupfs mounted accordingly; if set to
1638 "strict" a new cgroup namespace is allocated for the service, and
1639 cgroupfs is mounted read-only for the service.
1641 * The StateDirectory=, RuntimeDirectory=, CacheDirectory=,
1642 LogsDirectory=, and ConfigurationDirectory= settings gained support
1643 for configuring the respective directories as read-only, via a ':ro'
1644 flag that can be appended to each setting's value.
1646 * When DynamicUser= is combined with
1647 StateDirectory=/RuntimeDirectory=/CacheDirectory=/LogsDirectory= and
1648 ID mapped mounts are available on the referenced path, the data in
1649 there is now preferably made available by establishing ID mapped from
1650 the "nobody" user to the dynamic user, rather than via recursive
1653 * A new service property PrivatePIDs= has been added that runs executed
1654 processes as PID 1 - the init process - within their own PID
1655 namespace. PrivatePIDs= also mounts /proc/ so only processes within
1656 the new PID namespace are visible.
1660 * udev rules now set 'uaccess' for /dev/udmabuf, giving locally
1661 logged-in users access to the hardware. This is useful in order to
1662 support IPMI cameras with libcamera.
1664 * Serial port devices will no longer show up as systemd units, unless
1665 they have an IO port or memory assigned to them. This means that only
1666 serial ports that actually exist should show up as .device units now.
1668 * mtd devices (i.e. certain kinds of flash memory devices) will now
1669 show up as .device units in systemd.
1671 * The firmware_node/sun sysfs attribute will now be used (if available)
1672 for naming slot-based network interfaces, i.e. ID_NET_NAME_SLOT.
1673 Moreover the interface aliases specified in DeviceTree are now
1674 searched for both on the interface's parent device (as before) and
1675 the device itself (new).
1677 * Various USB hardware wallets are now recognized by udev via a .hwdb
1678 file, and get the ID_HARDWARE_WALLET= property set, which enables
1679 "uaccess" for them, i.e. direct unprivileged access.
1681 * udevadm info will now output the device ID string in lines prefixed
1682 with "J:", and the driver subsystem in lines prefixed with "B:".
1684 * udev rules files now support case-insensitive attribute matching
1685 (e.g. ATTR{foo}==i"abcd")
1689 * New DesignatedMaintenanceTime= configuration option allows shutdowns
1690 to be automatically scheduled at the specified time.
1692 * logind now reacts to Ctrl-Alt-Shift-Esc being pressed. It will send
1693 out a org.freedesktop.login1.SecureAttentionKey signal, indicating a
1694 request by the user for the system to display a secure login dialog.
1695 The handling of SAK can be suppressed in logind configuration.
1697 * logind now supports handing off session-managed access to hidraw
1698 devices via its D-Bus APIs, the same way it already supports that for
1699 DRM and evdev input devices. This permits unprivileged clients to get
1700 hidraw fds for a device, that are automatically suspended when the
1701 session switches away.
1703 * systemd-logind now exposes two D-Bus properties CanLock and CanIdle
1704 for all sessions. These properties indicate whether the session's
1705 class supports screen locking and idleness detection.
1707 * systemd-inhibit now allows interactive polkit authorization. It
1708 gained a --no-ask-password option to suppress it.
1712 * Unprivileged clients are now allowed to register VMs and containers.
1713 Machines started via the systemd-vmspawn@.service unit will now be
1714 registered with systemd-machined.
1716 * systemd-machined gained a pretty complete set of Varlink APIs
1717 exposing its functionality. This is an alternative to the
1718 pre-existing D-Bus interface.
1722 * The resolvconf command now supports '-p' switch. If specified, the
1723 interface will not be used as the default route for domain name
1726 * resolvectl now enables interactive polkit authorization. It gained a
1727 --no-ask-password option to suppress it.
1729 systemd-networkd and networkctl:
1731 * IPv6 address labels can be also configured in a new [IPv6AddressLabel]
1732 section with Prefix= and Label= settings in networkd.conf. Please see
1733 networkd.conf(5) for more details.
1735 * 'networkctl edit' can now read the new file contents from standard
1736 input with the new --stdin option.
1738 * 'networkctl edit' and 'cat' now support editing/showing .netdev files
1739 by link. 'networkctl cat' can also list all configuration files
1740 associated with an interface at once with ':all'.
1742 * networkctl gained a --no-ask-password option to suppress interactive
1743 polkit authorization.
1745 * "mac" has been added to the default AlternativeNamesPolicy= setting
1746 for network links (via 99-default.link). This means "enx*" interface
1747 names will now be added to the list of alternative interface names by
1748 default, for all interfaces that have a MAC address assigned
1751 * networkd .netdev bridge devices gained a new setting FDBMaxLearned=
1752 for setting a limit on the number of dynamically learned FDB entries.
1754 * networkd .network files for bridge devices now support Layer 2 (in
1755 addition to the pre-existing Layer 3) MDB entries, via
1756 MulticastGroupAddress=.
1758 * systemd-networkd will now log when per-network sysctls belonging to
1759 network interfaces managed by it are changed outside of networkd,
1760 thus highlighting conflict of ownership/management of these knobs.
1762 * systemd-networkd will now make RFC9463 DNR fields available to
1763 systemd-resolved, for automatic DNS DoT configuration, and similar.
1765 * The "dhcp" and "dhcp-on-stop" values for KeepConfiguration= setting in
1766 .network file are replaced with "dynamic" and "dynamic-on-stop",
1767 respectively. When specified, systemd-networkd will preserve all
1768 dynamic configurations via DHCPv4, DHCPv6, NDISC, and IPv4LL with
1769 ACD, while previously only DHCPv4 configurations were kept. Also,
1770 when systemd-networkd is restarted, regardless of the setting, these
1771 dynamic configurations are unconditionally kept. So, systemd-networkd
1772 can be restarted without disturbing ongoing connections.
1774 * systemd-networkd now updates traffic control configuration without
1775 clearing existing settings. Thus, those settings can be updated by
1776 editing relevant .network files and triggering 'networkctl reload'.
1778 * systemd-networkd now gracefully updates netdev settings specified in
1779 .netdev files when 'networkctl reload' is called. Previously, if the
1780 relevant interfaces existed, new settings would not be applied. Now,
1781 new settings will be applied if possible. Some settings cannot be
1782 updated after a netdev is configured, e.g. VLAN ID can be only
1783 specified on creation. To change such settings, user needs to remove
1784 existing interfaces, and invoke 'networkctl reload' or restart
1787 systemd-boot, systemd-stub, and related tools:
1789 * The EFI stub now supports loading of .ucode sections with microcode
1790 from PE add-on files. It also now supports loading .initrd sections
1791 from PE add-on files.
1793 * A new .profile PE section type is now documented and supported in
1794 systemd-measure, ukify, systemd-stub and systemd-boot. These new
1795 sections allow multiple "profiles" to be stored together in the UKI,
1796 where each .profile section creates groupings of sections in the UKI,
1797 allowing some sections to be shared and other sections like .cmdline
1798 or .initrd unique to the profile. This may be used to provide a
1799 single UKI that synthesizes multiple menu items in the boot menu (for
1800 example, a regular one to boot, plus a debugging one, or a factory
1801 reset one, and so on – which only differ in kernel command line, but
1804 * New .dtbauto and .hwids sections are now documented and supported in
1805 systemd-measure, ukify, systemd-stub, and systemd-boot. A single UKI
1806 can contain multiple .dtbauto sections, and the 'compatible' string
1807 therein will be compared with the equivalent field in the DTB
1808 provided by the firmware, if present. If absent, SMBIOS will be used
1809 to calculate hardware IDs (CHIDs) and look them up in the content of
1810 .hwids, hopefully revealing an fallback 'compatible' string. This
1811 allows including multiple DTBs in a single UKI, with systemd-stub
1812 automatically loading the correct one for the current hardware.
1814 * ukify gained an --extend switch to import an existing UKI to
1815 be extended, and a --measure-base= switch to support measurement
1816 of multi-profile UKIs.
1818 * ukify gained a --certificate-provider switch to use an OpenSSL
1819 provider to load the certificate used to sign artifacts, instead of
1820 having to provide the path to a file on disk.
1822 * bootctl, systemd-keyutil, systemd-measure, systemd-repart, and
1823 systemd-sbsign gained a new --certificate-source switch that allows
1824 loading the X.509 certificate from an OpenSSL provider instead of a
1827 * systemd-boot's menu will now react to volume up/down rocker presses
1828 the same way as to arrow up/down presses: they move the menu item up
1829 or down. This is useful on device form factors that have only a
1830 volume rocker but no arrow keys (e.g. phones).
1832 * systemd-stub will report the partition UUID and image identifier its
1833 UKI executable is placed on separately from the data systemd-boot
1834 provides about where to find its own executable, via EFI
1835 variables. This is useful when systemd-boot and UKIs are placed on
1836 distinct partitions (i.e. ESP and XBOOTLDR).
1838 * bootctl gained new switches --print-loader-path and --print-stub-path
1839 that output the path to the boot loader or UKI used for the current
1842 * bootctl kernel-identify now recognizes EFI add-ons.
1844 * bootctl gained a --random-seed=yes|no option to control provisioning
1845 of the random seed file in the ESP. (This is useful when producing an
1846 image that will be used in multiple instances.)
1848 * bootctl now optionally supports installing UEFI Secure Boot databases
1849 (i.e. db/dbx/… databases in ESL format) for systemd-boot to pick up
1850 and automatically enroll if the system is booted in Setup Mode. This
1851 is controlled via bootctl's new --secure-boot-auto-enroll=yes switch
1852 (and some auxiliary ones). A certificate can be provided in DER
1853 format, and is automatically converted into an ESL, as needed.
1855 * bootctl, systemd-measure, systemd-repart when referencing signing
1856 keys on OpenSSL engines may now query for PINs and similar via
1857 systemd's native systemd-ask-password logic (and take benefit of its
1860 * A new systemd-sbsign tool has been added, that can be used to sign
1861 EFI binaries (PE) for Secure Boot. This tool supports OpenSSL engines
1862 and providers, with pin caching support for PKCS11. ukify supports it
1863 as an alternative to sbsigntool and pesign.
1865 * A new systemd-keyutil tool has been added, that can be used to perform
1866 various operations on private keys and X.509 certificates.
1870 * journalctl can now list invocations of a unit with the
1871 --list-invocation options and show logs for a specific invocation
1872 with the new --invocation/-I option. (This is analogous to the
1873 --list-boots/--boot/-b options.)
1875 systemd-sysupdate and related tools:
1877 * systemd-sysupdated has been added as system service, allowing
1878 unprivileged clients to update the system via D-Bus calls. Note that
1879 for now the systemd-sysupdated API is considered experimental, and is
1880 not considered stable yet.
1882 A new updatectl command-line tool can be used to control the
1885 * systemd-sysupdate gained a new --offline option to force it to
1886 operate locally. This is useful when listing locally installed
1889 * systemd-sysupdate gained a new --transfer-source= option to set the
1890 directory to which transfer sources configured with
1891 PathRelativeTo=explicit will be interpreted.
1893 * systemd-sysupdate now reports download progress via sd_notify().
1895 * systemd-sysupdate now supports output in JSON mode for all commands.
1897 * systemd-sysupdate definitions may now carry references to ChangeLog
1898 and AppStream metadata.
1900 * Transfer definitions for systemd-sysupdate are supposed to carry the
1901 ".transfer" suffix now, changing from ".conf". The latter remains
1902 supported for compatibility, but it's recommended to rename all files
1903 reflecting this suffix change.
1905 * systemd-sysupdate now supports new ".feature" files that may be
1906 used in conjunction with ".transfer" files to group them together, and
1907 allow them to be turned off or on, individually per group.
1909 TPM & systemd-cryptsetup:
1911 * The 'has-tpm2' verb which reports whether TPM2 functionality is
1912 available has been moved from systemd-creds to systemd-analyze.
1914 * systemd-tpm2-setup will gracefully handle TPMs that have a PIN set on
1915 the TPM, and not attempt to automatically set up a Storage Root Key
1918 * New crypttab option password-cache=yes|no|read-only can be used to
1919 customize password caching.
1921 * New crypttab options fido2-pin=, fido2-up=, fido2-uv= can be used to
1922 enable/disable the PIN query, User Presence check, and User
1925 * systemd-cryptenroll gained new options --fido2-salt-file= and
1926 --fido2-parameters-in-header= to simplify manual enrollment of FIDO2
1929 * systemd-cryptenroll, systemd-repart, and systemd-storagetm gained a
1930 new --list-devices option to list appropriate candidate block
1933 * systemd-cryptenroll/systemd-cryptsetup now support combined signed
1934 PCR policies and local systemd-pcrlock policies for unlocking a
1935 disk. Or in other words, it's now possible to bind unlocking of a
1936 local disk to a specific OS vendor *and* a locally managed set of
1937 measurements describing the local system.
1941 * varlinkctl gained a new verb 'list-methods' to show a list of
1942 methods implemented by a service.
1944 * varlinkctl gained a --quiet/-q option to suppress method call
1947 * varlinkctl gained a --graceful= option to suppress specific Varlink
1948 errors, and treat them as success.
1950 * varlinkctl gained a --timeout= option to limit how long the
1951 invocation can take.
1953 * varlinkctl allows remote invocations over ssh, via the new
1954 "ssh-exec:" address specification. It'll make an ssh connection,
1955 start the specified executable on the remote side, and communicate
1956 with the remote process using the Varlink protocol.
1958 The "ssh:" address specification has been renamed to "ssh-unix:"
1959 (reflecting the fact it is used to connect to a remote AF_UNIX socket
1960 via SSH). The old syntax is still supported for backwards
1963 * varlinkctl's 'introspect' verb no longer requires specification of an
1964 interface name. If none is specified all interfaces exposed by the
1965 service are shown. Moreover, more than one interface name may be
1966 specified now, in which case all specified ones are displayed.
1970 * systemd-repart's CopyBlocks= directive can now use a character device
1971 as source (in addition to previously supported regular files and
1972 block devices). This is useful for initializing a partition from
1973 /dev/urandom or similar.
1975 * systemd-repart gained new Compression= and CompressionLevel= settings
1976 to enable internal compression in filesystems created offline.
1978 * systemd-repart understands a new MakeSymlinks= option to create one
1979 or more symlinks (each specified as a symlink name and target) within
1980 a newly formatted file system.
1982 * systemd-repart gained a new SupplementFor= setting that allows
1983 allocating a partition only if some other existing partition cannot
1984 be adjusted to match the constraints defined for it. This is useful
1985 to generate an XBOOTLDR partition if and only if an ESP already
1986 exists that is too small for the required constraints.
1988 * The default size of verity hash partitions is now automatically
1989 derived from SizeMaxBytes= of the data partition it is protecting.
1993 * systemd-ssh-proxy now also supports the AF_UNIX-based "VSOCK MUX"
1994 protocol used by CloudHypervisor/Firecracker to expose AF_VSOCK
1995 sockets of the VM on the host. Or in other words: it's now possible
1996 to directly connect to ssh via AF_VSOCK from hosts to VMs of these
1997 two hypervisors (previously this was only supported for hypervisors
1998 which expose AF_VSOCK on the host as AF_VSOCK, such as qemu).
2000 * systemd-ssh-proxy can now reference local VMs by their name: connect
2001 to any local VM "foobar" registered with systemd-machined via "ssh
2002 machine/foobar" using the AF_VSOCK protocol.
2006 * systemd-analyze will now show the SMBIOS #11 vendor strings set for
2007 the machine with a new 'smbios11' verb.
2009 * systemd-analyze gained a new --instance= option that can be used to
2010 provide an instance name to analyze multiple templates instantiated
2011 with the same instance name.
2013 * systemd-analyze's "capability" verb now gained a new --mask
2014 parameter. If specified a numeric capbality mask can be specified
2015 which is decoded for its contained capabilities.
2017 * systemd-analyze's "plot" verb gained two new settings: --scale-svg=
2018 allows the X axis of the split to be stritched by a factor. If
2019 --detailed is specified activation timestamps are shown in the plot.
2023 * 'busctl monitor' gained new options --limit-messages= and --timeout=
2024 to set the number of matches or limit the runtime of the command.
2026 * busctl now supports doing method calls with embedded unix file
2029 * busctl acquired a new "wait" command to wait for a specific signal to
2034 * systemd-nspawn --bind-user= will now propagate the bound user's SSH
2035 public key (if included in the user record) into the container,
2036 ensuring that any such bound user is directly accessible via ssh.
2038 * systemd-nspawn now supports unprivileged FUSE inside containers.
2042 * A new generator sytemd-import-generator has been added to synthesize
2043 image download jobs. This provides functionality similar to
2044 importctl, but is configured via the kernel command line and system
2045 credentials. It may be used to automatically download sysext,
2046 confext, portable service, nspawn container or vmspawn VM images at
2049 * systemd-importd now provides a Varlink IPC interface, in addition to
2050 its existing D-Bus IPC interface.
2052 * The individual import/export tools will now display a nice progress
2053 bar when downloading files.
2055 systemd-userdb & systemd-homed:
2057 * userdbctl gained a pair of switches --uid-min= and --uid-max= to
2058 filter the UID/GID range of the listed users or groups. It also
2059 gained a new switch --disposition= to filter them by disposition
2060 (i.e. show only system users or only regular users, and so on). It
2061 also gained a new switch --fuzzy that permits a "fuzzy" search for a
2062 user, i.e. doing a substring and string distance search, and looking
2063 into the real name field of the user and other similar fields. It
2064 gained a new switch --boundaries=no for disabling display of the
2065 UID/GID range boundaries in its output.
2067 * User records learnt a new set of fields that may list field names
2068 that may be changed by the user themselves without requiring
2069 administrator authentication. This new field is honoured by
2070 systemd-homed to allow users to change selected properties of their
2075 * run0 gained a new pair of settings --pty and --pipe that control
2076 whether to invoke the specified binary on a freshly allocated pseudo
2077 TTY, or whether to pass the client's STDIN/STDOUT/STDERR through
2080 * run0 gained a new switch --shell-prompt-prefix= that permits passing
2081 in a string to display on each shell prompt as prefix. If not
2082 specified otherwise this will show a superhero emoji (🦸), in order
2083 to visually communicate the temporarily elevated privileges a run0
2084 session provides. This makes use of the $SHELL_PROMPT_PREFIX
2085 environment variables mentioned below.
2087 * systemd-run can output some of its runtime data in JSON format via
2088 the new --json= option.
2092 * systemd-tmpfiles --purge switch now requires specification of at
2093 least one tmpfiles.d/ drop-in file.
2095 * tmpfiles.d/ files gained a new '?' specifier for the 'L' line type to
2096 create a symlink only if the source exists, and gracefully skip the
2101 * systemctl now supports the --now option with the 'reenable' verb.
2103 * systemd-mount can now output JSON with a new --json= switch, for use
2104 with --list-devices. It also shows the "diskseq" property in the
2107 * systemd-id128 gained a new 'var-partition-uuid' verb to calculate
2108 the DPS UUID for /var/ keyed by the local machine-id.
2110 * localectl gained a -l/--full option to show output without
2113 * timedatectl now supports interactive polkit authorization.
2115 * The new Linux mseal(), listmount(), statmount() syscalls have been
2116 added to relevant system call groups.
2118 * The systemd-ask-password logic has been extended with a per-user
2119 scope, i.e. user programs may now ask for passwords via the same
2120 mechanism and the previously system-wide only mechanism.
2122 * A new set of system/service credentials are added:
2123 shell.prompt.prefix, shell.prompt.suffix and shell.welcome. At login
2124 time these are propagated into the $SHELL_PROMPT_PREFIX,
2125 $SHELL_PROMPT_SUFFIX, $SHELL_PROMPT_WELCOME environment
2126 variables. These in turn are included in the shell prompt of
2127 interactive shells and shown at login time, via
2128 /etc/profile.d/70-systemd-shell-extra.sh. This functionality is
2129 useful to visually highlight the fact a specific shell prompt
2130 originates from a specific system, execution context or tool. These
2131 credentials and environment variables are supposed to be generically
2132 useful within and outside of the immediate systemd context. It is
2133 also used by 'run0', see above.
2135 * New RELEASE_TYPE=, EXPERIMENT=, EXPERIMENT_URL= fields have been
2136 defined for the /etc/os-release file. For example,
2137 "RELEASE_TYPE=development|stable|lts" can be used to indicate various
2138 stages of the release life cycle, and "RELEASE_TYPE=experimental" can
2139 indicate experimental builds, with the EXPERIMENT= field providing a
2140 human-readable description of the nature of the experiment.
2142 * A new sleep.conf HibernateOnACPower= option has been added, which
2143 when disabled will suppress hibernation in suspend-then-hibernate
2144 mode until the system is disconnected from a power source.
2146 * A bunch of patches to ease building against musl have been merged.
2148 * The various components that display progress bars
2149 (i.e. systemd-repart, systemd-sysupdate/updatectl, importctl), will
2150 now also issue the ANSI sequences for progress reports that Windows
2151 Terminal understands. Most Linux terminals currently do not support
2152 this sequence (and ignore it), but hopefully this will change one
2153 day. The progress information is used to display a nice progress
2154 animation in the terminal tab and icon. For details about the ANSI
2155 sequence and its effects, see:
2157 https://github.com/microsoft/terminal/pull/8055
2158 https://conemu.github.io/en/AnsiEscapeCodes.html#ConEmu_specific_OSC
2160 * systemd-sysusers is now able to create fully locked user
2161 accounts. For compatibility it so far created accounts with a locked
2162 (i.e. invalid) password, but not marked locked as a whole. With the
2163 new "!" modifier for "u" lines, it is now possible to create fully
2164 locked accounts. The distinction between accounts with a locked
2165 password and fully locked accounts is relevant when considering
2166 non-password forms of authentication, i.e. SSH and such. It is
2167 strongly recommended to make use of this new feature for almost all
2168 system accounts, since they usually do not require (and should not
2169 permit) interactive logins. All of systemd's own system users have
2170 been changed to be marked as fully locked.
2172 * systemd-coredump now supports a new EnterNamespace= option, which
2173 defaults to off. If enabled systemd-coredump will access the mount
2174 namespace of any crashed process to acquire debug symbol information,
2175 in order to be able to symbolize backtraces. This option is useful to
2176 improve backtraces of processes of containerized applications. (Note
2177 that the host systemd-coredump preferably dispatches coredump
2178 processing to the container itself, if it supports that. Only full-OS
2179 containers which run systemd inside will support this however, in
2180 other cases EnterNamespace= might be an suitable approach to acquire
2181 symbolized backtraces.)
2183 Special thanks to Nick Owens for bringing attention to and testing
2184 fixes for issue #34516.
2186 Contributions from: 12paper, A. Wilcox, Abderrahim Kitouni,
2187 Adrian Vovk, Alain Greppin, Allison Karlitskaya, Alyssa Ross,
2188 Anders Jonsson, Andika Triwidada, Andreas Schwab, Andres Beltran,
2189 Ani Sinha, Anouk Ceyssens, Anselm Schueler, Anton Golubev,
2190 Antonio Alvarez Feijoo, Arian van Putten, Arnaud Patard,
2191 Arthur Shau, Bastien Nocera, Benjamin ROBIN, Brenton Simpson,
2192 Bryan Gurney, ButterflyOfFire, Carlo Teubner, Celeste Liu,
2193 Chen Guanqiao, Chen Qi, Chengen Du, Christian Hesse,
2194 Christoph Anton Mitterer, Colin Foster, Collin L,
2195 Cristian Rodríguez, Daan De Meyer, Dan Nicholson, Daniel Dawson,
2196 Daniel Martinez, Daniel P. Berrangé, Daniel Rusek,
2197 Darsey Litzenberger, David Joaquín Shourabi Porcel,
2198 David Michael, David Rheinsberg, David Tardon, Davide Cavalca,
2199 Derek J. Clark, Diego Viola, Dimitrys Meliates, Diogo Ivo,
2200 Dmytro Markevych, DocNITE, Dominique Martinet,
2201 Dr. David Alan Gilbert, Edson Juliano Drosdeck, Erik Sjölund,
2202 Etienne Champetier, Etienne Cordonnier, Ettore Atalan,
2203 Eugeny Shcheglov, Excited-bore, Fabian Vogt, Federico Giovanardi,
2204 Filip Lewiński, Florian Schmaus, Franck Bui, Frantisek Sumsal,
2205 Fábio Rodrigues Ribeiro, Gabriel Elyas, Gaël PORTAY,
2206 Geraldo S. Simião Kutz, Giovanni Baratta, Greg Heartsfield,
2207 Gregor Herburger, Gregory Arenius, GwynBleidD, Göran Uddeborg,
2208 Hans de Goede, Helmut Grohne, Henry Chen, Ian Abbott, Integral,
2209 Ivan Kruglov, Ivan Shapovalov, James Coglan, James Hilliard,
2210 James Muir, Jason Yundt, Jeffrey Bosboom, Jian Zhang,
2211 Jiri Grönroos, Johannes Schneider, John A. Leuenhagen,
2212 Jose Ignacio Tornos Martinez, JoseskVolpe, Joshua Grisham,
2213 Jörg Behrmann, Kai-Chuan Hsieh, Kamil Szczęk, Karel Zak,
2214 Kornilios Kourtis, Kuntal Majumder, Lennart Poettering,
2215 Lidong Zhong, Luca Boccassi, Lucas Adriano Salles,
2216 Lucas Werkmeister, Ludwig Nussel, Luke T. Shumaker,
2217 Lukáš Nykrýn, Luna Jernberg, Léane GRASSER, Maanya Goenka,
2218 Mantas Mikulėnas, Marc Reisner, Marcel Hellwig, Marco Tomaschett,
2219 Marin Kresic, Marius Hoch, Martin Srebotnjak, Martin Wilck,
2220 Mary Strodl, Matteo Croce, Matthias Lisin, Matthias Schiffer,
2221 Matthieu Baerts (NGI0), Matthieu CHARETTE,
2222 Mauri de Souza Meneguzzo, Maximilian Wilhelm, Merlin Jehli,
2223 Michael Ferrari, Michal Koutný, Michal Sekletár, Michał Górny,
2224 Michele Dionisio, Michiel, Mickaël Salaün, Mike Gilbert,
2225 Mike Yuan, MkKvcs, Nick Cao, Nick Rosbrook, Nils K, Nova840,
2226 Oğuz Ersen, Pavel Borecki, PavlNekrasov, Peter Hutterer,
2227 Peter Rajnoha, Piotr Drąg, Raphaël Mélotte, Renan Guilherme,
2228 Renjaya Raga Zenta, Ricky Tigg, Riku, Robin Lee, Ronan Pigott,
2229 Ryan Wilson, Sam James, Sascha Mester, Sean Rhodes,
2230 Sebastian Gross, Septatrix, Sergey A, ShreyasMahangade,
2231 Simon Pilkington, Skye Chappelle, Steve Traylen, Stuart Hayhurst,
2232 SuhailAhmedVelorum, Susant Sahani, Takeo Kondo, Temuri Doghonadze,
2233 Thomas Blume, Thorsten Kukuk, Thorsten Scherer, Tobias Fleig,
2234 Tobias Zimmermann, Tom Coldrick, Tom Yan, Tomas Bzatek,
2235 Topi Miettinen, Tristan F.-R., Uday Shankar, Valentin David,
2236 Vasiliy Kovalev, Vitaly Kuznetsov, Vito Caputo, Vladimir Panteleev,
2237 Vursc, Will Fancher, WilliButz, Winterhuman, Xeonacid, Xuanjun Wen,
2238 Yanqing Jing, Yaron Shahrabani, Yu Watanabe, Yuri Chornoivan,
2239 ZHANG Yuntian, Zbigniew Jędrzejewski-Szmek, Zhou Qiankang,
2240 andre4ik3, anonymix007, bryango, chayleaf, chenjiayi, csp5me, cvlc12,
2241 fwfy, gerblesh, hugo303, jan@neighbourhood.ie, jauge-technica,
2242 lumingzh, maia x., marginaldev, migleeson, nerdopolis, oldherl,
2243 pyfisch, q66, rajmohan r, reDBo0n, rhellstrom, rindeal, samuelvw01,
2244 sinus-x, tfg13, vdovhanych, xujing, Łukasz Stelmach,
2245 Štěpán Němec, Дамјан Георгиевски, 白一百
2247 — Edinburgh, 2024-12-10
2251 Announcements of Future Feature Removals and Incompatible Changes:
2253 * Support for automatic flushing of the nscd user/group database caches
2254 will be dropped in a future release.
2256 * Support for cgroup v1 ('legacy' and 'hybrid' hierarchies) is now
2257 considered obsolete and systemd by default will refuse to boot under
2258 it. To forcibly reenable cgroup v1 support,
2259 SYSTEMD_CGROUP_ENABLE_LEGACY_FORCE=1 must be set on kernel command
2260 line. The meson option 'default-hierarchy=' is also deprecated, i.e.
2261 only cgroup v2 ('unified' hierarchy) can be selected as build-time
2264 * Support for System V service scripts is deprecated and will be
2265 removed in a future release. Please make sure to update your software
2266 *now* to include a native systemd unit file instead of a legacy
2267 System V script to retain compatibility with future systemd releases.
2269 * Support for the SystemdOptions EFI variable is deprecated.
2270 'bootctl systemd-efi-options' will emit a warning when used. It seems
2271 that this feature is little-used and it is better to use alternative
2272 approaches like credentials and confexts. The plan is to drop support
2273 altogether at a later point, but this might be revisited based on
2276 * systemd-run's switch --expand-environment= which currently is disabled
2277 by default when combined with --scope, will be changed in a future
2278 release to be enabled by default.
2280 * Previously, systemd-networkd did not explicitly remove any bridge
2281 VLAN IDs assigned on bridge master and ports. Since version 256, if a
2282 .network file for an interface has at least one valid setting in the
2283 [BridgeVLAN] section, then all assigned VLAN IDs on the interface
2284 that are not configured in the .network file are removed.
2286 * IPForward= setting in .network file is deprecated and replaced with
2287 IPv4Forwarding= and IPv6Forwarding= settings. These new settings are
2288 supported both in .network file and networkd.conf. If specified in a
2289 .network file, they control corresponding per-link settings. If
2290 specified in networkd.conf, they control corresponding global
2291 settings. Note, previously IPv6SendRA= and IPMasquerade= implied
2292 IPForward=, but now they imply the new per-link settings. One of the
2293 simplest ways to migrate configurations, that worked as a router with
2294 the previous version, is enabling both IPv4Forwarding= and
2295 IPv6Forwarding= in networkd.conf. See systemd.network(5) and
2296 networkd.conf(5) for more details.
2298 * systemd-gpt-auto-generator will stop generating units for ESP or
2299 XBOOTLDR partitions if it finds mount entries for or below the /boot/
2300 or /efi/ hierarchies in /etc/fstab. This is to prevent the generator
2301 from interfering with systems where the ESP is explicitly configured
2302 to be mounted at some path, for example /boot/efi/ (this type of
2303 setup is obsolete, but still commonly found).
2305 * The behavior of systemd-sleep and systemd-homed has been updated to
2306 freeze user sessions when entering the various sleep modes or when
2307 locking a homed-managed home area. This is known to cause issues with
2308 the proprietary NVIDIA drivers. Packagers of the NVIDIA proprietary
2309 drivers may want to add drop-in configuration files that set
2310 SYSTEMD_SLEEP_FREEZE_USER_SESSIONS=false for systemd-suspend.service
2311 and related services, and SYSTEMD_HOME_LOCK_FREEZE_SESSION=false for
2312 systemd-homed.service.
2314 * systemd-tmpfiles and systemd-sysusers, when given a relative
2315 configuration file path (with at least one directory separator '/'),
2316 will open the file directly, instead of searching for the given
2317 partial path in the standard locations. The old mode wasn't useful
2318 because tmpfiles.d/ and sysusers.d/ configuration has a flat
2319 structure with no subdirectories under the standard locations and
2320 this change makes it easier to work with local files with those
2323 * systemd-tmpfiles now properly applies nested configuration to 'R' and
2324 'D' stanzas. For example, with the combination of 'R /foo' and 'x
2325 /foo/bar', /foo/bar will now be excluded from removal.
2327 * systemd.crash_reboot and related settings are deprecated in favor of
2328 systemd.crash_action=.
2330 * Stable releases for version v256 and newer will now be pushed in the
2331 main repository. The systemd-stable repository will be used for existing
2332 stable branches (v255-stable and lower), and when they reach EOL it will
2335 General Changes and New Features:
2337 * Various programs will now attempt to load the main configuration file
2338 from locations below /usr/lib/, /usr/local/lib/, and /run/, not just
2339 below /etc/. For example, systemd-logind will look for
2340 /etc/systemd/logind.conf, /run/systemd/logind.conf,
2341 /usr/local/lib/systemd/logind.conf, and /usr/lib/systemd/logind.conf,
2342 and use the first file that is found. This means that the search
2343 logic for the main config file and for drop-ins is now the same.
2345 Similarly, kernel-install will look for the config files in
2346 /usr/lib/kernel/ and the other search locations, and now also
2349 systemd-udevd now supports drop-ins for udev.conf.
2351 * A new 'systemd-vpick' binary has been added. It implements the new
2352 vpick protocol, where a "*.v/" directory may contain multiple files
2353 which have versions (following the UAPI version format specification)
2354 embedded in the file name. The files are ordered by version and
2355 the newest one is selected.
2357 systemd-nspawn --image=/--directory=, systemd-dissect,
2358 systemd-portabled, and the RootDirectory=, RootImage=,
2359 ExtensionImages=, and ExtensionDirectories= settings for units now
2360 support the vpick protocol and allow the latest version to be
2361 selected automatically if a "*.v/" directory is specified as the
2364 * Encrypted service credentials can now be made accessible to
2365 unprivileged users. systemd-creds gained new options --user/--uid=
2366 for encrypting/decrypting a credential for a specific user.
2368 * New command-line tool 'importctl' to download, import, and export
2369 disk images via systemd-importd is added with the following verbs:
2370 pull-tar, pull-raw, import-tar, import-raw, import-fs, export-tar,
2371 export-raw, list-transfers, and cancel-transfer. This functionality
2372 was previously available in "machinectl", where it was used
2373 exclusively for machine images. The new "importctl" generalizes this
2374 for sysext, confext, and portable service images.
2376 * The systemd sources may now be compiled cleanly with all OpenSSL 3.0
2377 deprecations removed, including the OpenSSL engine logic turned off.
2381 * New system manager setting ProtectSystem= has been added. It is
2382 analogous to the unit setting, but applies to the whole system. It is
2383 enabled by default in the initrd.
2385 Note that this means that code executed in the initrd cannot naively
2386 expect to be able to write to /usr/ during boot. This affects
2387 dracut <= 101, which wrote "hooks" to /lib/dracut/hooks/. See
2388 https://github.com/dracut-ng/dracut-ng/commit/a45048b80c27ee5a45a380.
2390 * New unit setting WantsMountsFor= has been added. It is analogous to
2391 RequiresMountsFor=, but creates a Wants= dependency instead of
2392 Requires=. This new logic is now used in various places where mounts
2393 were added as dependencies for other settings (WorkingDirectory=-…,
2394 PrivateTmp=yes, cryptsetup lines with 'nofail').
2396 * New unit setting MemoryZSwapWriteback= can be used to control the new
2397 memory.zswap.writeback cgroup knob added in kernel 6.8.
2399 * The manager gained a org.freedesktop.systemd1.StartAuxiliaryScope()
2400 D-Bus method to devolve some processes from a service into a new
2401 scope. This new scope will remain running, even when the original
2402 service unit is restarted or stopped. This allows a service unit to
2403 split out some worker processes which need to continue running.
2404 Control group properties of the new scope are copied from the
2405 originating unit, so various limits are retained.
2407 * Units now expose properties EffectiveMemoryMax=,
2408 EffectiveMemoryHigh=, and EffectiveTasksMax=, which report the
2409 most stringent limit systemd is aware of for the given unit.
2411 * A new unit file specifier %D expands to $XDG_DATA_HOME (for user
2412 services) or /usr/share/ (for system services).
2414 * AllowedCPUs= now supports specifier expansion.
2416 * What= setting in .mount and .swap units now accepts fstab-style
2417 identifiers, for example UUID=… or LABEL=….
2419 * RestrictNetworkInterfaces= now supports alternative network interface
2422 * PAMName= now implies SetLoginEnvironment=yes.
2424 * systemd.firstboot=no can be used on the kernel command-line to
2425 disable interactive queries, but allow other first boot configuration
2426 to happen based on credentials.
2428 * The system's hostname can be configured via the systemd.hostname
2431 * The systemd binary will no longer chainload sysvinit's "telinit"
2432 binary when called under the init/telinit name on a system that isn't
2433 booted with systemd. This previously has been supported to make sure
2434 a distribution that has both init systems installed can reasonably
2435 switch from one to the other via a simple reboot. Distributions
2436 apparently have lost interest in this, and the functionality has not
2437 been supported on the primary distribution this was still intended
2438 for a long time, and hence has been removed now.
2440 * A new concept called "capsules" has been introduced. "Capsules" wrap
2441 additional per-user service managers, whose users are transient and
2442 are only defined as long as the service manager is running. (This is
2443 implemented via DynamicUser=1), allowing a user manager to be used to
2444 manage a group of processes without needing to create an actual user
2445 account. These service managers run with home directories of
2446 /var/lib/capsules/<capsule-name> and can contain regular services and
2447 other units. A capsule is started via a simple "systemctl start
2448 capsule@<name>.service". See the capsule@.service(5) man page for
2451 Various systemd tools (including, and most importantly, systemctl and
2452 systemd-run) have been updated to interact with capsules via the new
2453 "--capsule="/"-C" switch.
2455 * .socket units gained a new setting PassFileDescriptorsToExec=, taking
2456 a boolean value. If set to true the file descriptors the socket unit
2457 encapsulates are passed to the ExecStartPost=, ExecStopPre=,
2458 ExecStopPost= using the usual $LISTEN_FDS interface. This may be used
2459 for doing additional initializations on the sockets once they are
2460 allocated. (For example, to install an additional eBPF program on
2463 * The .socket setting MaxConnectionsPerSource= (which so far put a
2464 limit on concurrent connections per IP in Accept=yes socket units),
2465 now also has an effect on AF_UNIX sockets: it will put a limit on the
2466 number of simultaneous connections from the same source UID (as
2467 determined via SO_PEERCRED). This is useful for implementing IPC
2468 services in a simple Accept=yes mode.
2470 * The service manager will now maintain a counter of soft reboot cycles
2471 the system went through. It may be queried via the D-Bus APIs.
2473 * systemd's execution logic now supports the new pidfd_spawn() API
2474 introduced by glibc 2.39, which allows us to invoke a subprocess in a
2475 target cgroup and get a pidfd back in a single operation.
2477 * systemd/PID 1 will now send an additional sd_notify() message to its
2478 supervising VMM or container manager reporting the selected hostname
2479 ("X_SYSTEMD_HOSTNAME=") and machine ID ("X_SYSTEMD_MACHINE_ID=") at
2480 boot. Moreover, the service manager will send additional sd_notify()
2481 messages ("X_SYSTEMD_UNIT_ACTIVE=") whenever a target unit is
2482 reached. This can be used by VMMs/container managers to schedule
2483 access to the system precisely. For example, the moment a system
2484 reports "ssh-access.target" being reached a VMM/container manager
2485 knows it can now connect to the system via SSH. Finally, a new
2486 sd_notify() message ("X_SYSTEMD_SIGNALS_LEVEL=2") is sent the moment
2487 PID 1 has successfully completed installation of its various UNIX
2488 process signal handlers (i.e. the moment where SIGRTMIN+4 sent to
2489 PID 1 will start to have the effect of shutting down the system
2490 cleanly). X_SYSTEMD_SHUTDOWN= is sent shortly before the system shuts
2491 down, and carries a string identifying the type of shutdown,
2492 i.e. "poweroff", "halt", "reboot". X_SYSTEMD_REBOOT_PARAMETER= is
2493 sent at the same time and carries the string passed to "systemctl
2494 --reboot-argument=" if there was one.
2496 * New D-Bus properties ExecMainHandoffTimestamp and
2497 ExecMainHandoffTimestampMonotonic are now published by services
2498 units. This timestamp is taken as the very last operation before
2499 handing off control to invoked binaries. This information is
2500 available for other unit types that fork off processes (i.e. mount,
2501 swap, socket units), but currently only via "systemd-analyze dump".
2503 * An additional timestamp is now taken by the service manager when a
2504 system shutdown operation is initiated. It can be queried via D-Bus
2505 during the shutdown phase. It's passed to the following service
2506 manager invocation on soft reboots, which will then use it to log the
2507 overall "grey-out" time of the soft reboot operation, i.e. the time
2508 when the shutdown began until the system is fully up again.
2510 * "systemctl status" will now display the invocation ID in its usual
2511 output, i.e. the 128bit ID uniquely assigned to the current runtime
2512 cycle of the unit. The ID has been supported for a long time, but is
2513 now more prominently displayed, as it is a very useful handle to a
2514 specific invocation of a service.
2516 * systemd now generates a new "taint" string "unmerged-bin" for systems
2517 that have /usr/bin/ and /usr/sbin/ separate. It's generally
2518 recommended to make the latter a symlink to the former these days.
2520 * A new systemd.crash_action= kernel command line option has been added
2521 that configures what to do after the system manager (PID 1) crashes.
2522 This can also be configured through CrashAction= in systemd.conf.
2524 * "systemctl kill" now supports --wait which will make the command wait
2525 until the signalled services terminate.
2529 * systemd-journald can now forward journal entries to a socket
2530 (AF_INET, AF_INET6, AF_UNIX, or AF_VSOCK). The socket can be
2531 specified in journald.conf via a new option ForwardToSocket= or via
2532 the 'journald.forward_to_socket' credential. Log records are sent in
2533 the Journal Export Format. A related setting MaxLevelSocket= has been
2534 added to control the maximum log levels for the messages sent to this
2537 * systemd-journald now also reads the journal.storage credential when
2538 determining where to store journal files.
2540 * systemd-vmspawn gained a new --forward-journal= option to forward the
2541 virtual machine's journal entries to the host. This is done over a
2542 AF_VSOCK socket, i.e. it does not require networking in the guest.
2544 * journalctl gained option '-i' as a shortcut for --file=.
2546 * journalctl gained a new -T/--exclude-identifier= option to filter
2547 out certain syslog identifiers.
2549 * journalctl gained a new --list-namespaces option.
2551 * systemd-journal-remote now also accepts AF_VSOCK and AF_UNIX sockets
2552 (so it can be used to receive entries forwarded by systemd-journald).
2554 * systemd-journal-gatewayd allows restricting the time range of
2555 retrieved entries with a new "realtime=[<since>]:[<until>]" URL
2558 * systemd-cat gained a new option --namespace= to specify the target
2559 journal namespace to which the output shall be connected.
2561 * systemd-bsod gained a new option --tty= to specify the output TTY
2565 * /dev/ now contains symlinks that combine by-path and by-{label,uuid}
2568 /dev/disk/by-path/<path>/by-<label|uuid|…>/<label|uuid|…>
2570 This allows distinguishing partitions with identical contents on
2571 multiple storage devices. This is useful, for example, when copying
2572 raw disk contents between devices.
2574 * systemd-udevd now creates persistent /dev/media/by-path/ symlinks for
2575 media controllers. For example, the uvcvideo driver may create
2576 /dev/media0 which will be linked as
2577 /dev/media/by-path/pci-0000:04:00.3-usb-0:1:1.0-media-controller.
2579 * A new unit systemd-udev-load-credentials.service has been added
2580 to pick up udev.conf drop-ins and udev rules from credentials.
2582 * 'udevadm test' and 'udevadm test-builtin' commands now do not change
2583 any settings; sysfs attributes, sysctls, udev database and so on.
2584 E.g. 'udevadm test-builtin net_setup_link /sys/class/net/INTERFACE'
2585 does not change any interface settings, but only prints which .link
2586 file matches the interface. So, even privileged users can safely
2587 invoke the commands.
2589 * An allowlist/denylist may be specified to filter which sysfs
2590 attributes are used when crafting network interface names. Those
2591 lists are stored as hwdb entries
2592 ID_NET_NAME_ALLOW_<sysfsattr>=0|1
2594 ID_NET_NAME_ALLOW=0|1.
2596 The goal is to avoid unexpected changes to interface names when the
2597 kernel is updated and new sysfs attributes become visible.
2599 * A new unit tpm2.target has been added to provide a synchronization
2600 point for units which expect the TPM hardware to be available. A new
2601 generator "systemd-tpm2-generator" has been added that will insert
2602 this target whenever it detects that the firmware has initialized a
2603 TPM, but Linux hasn't loaded a driver for it yet.
2605 * systemd-backlight now properly supports numbered devices which the
2606 kernel creates to avoid collisions in the leds subsystem.
2608 * systemd-hwdb update operation can be disabled with a new environment
2609 variable SYSTEMD_HWDB_UPDATE_BYPASS=1.
2613 * systemd-hostnamed now exposes the machine ID and boot ID via
2614 D-Bus. It also exposes the hosts AF_VSOCK CID, if available.
2616 * systemd-hostnamed now provides a basic Varlink interface.
2618 * systemd-hostnamed exports the full data in os-release(5) and
2619 machine-info(5) via D-Bus and Varlink.
2621 * hostnamectl now shows the system's product UUID and hardware serial
2626 * systemd-networkd now provides a basic Varlink interface.
2628 * systemd-networkd's ARP proxy support gained a new option to configure
2629 a private VLAN variant of the proxy ARP supported by the kernel under
2630 the name IPv4ProxyARPPrivateVLAN=.
2632 * systemd-networkd now exports the NamespaceId and NamespaceNSID
2633 properties via D-Bus and Varlink. (which expose the inode and NSID of
2634 the network namespace the networkd instance manages)
2636 * systemd-networkd now supports IPv6RetransmissionTimeSec= and
2637 UseRetransmissionTime= settings in .network files to configure
2638 retransmission time for IPv6 neighbor solicitation messages.
2640 * networkctl gained new verbs 'mask' and 'unmask' for masking networkd
2641 configuration files such as .network files.
2643 * 'networkctl edit --runtime' allows editing volatile configuration
2644 under /run/systemd/network/.
2646 * The implementation behind TTLPropagate= network setting has been
2647 removed and the setting is now ignored.
2649 * systemd-network-generator will now pick up .netdev/.link/.network/
2650 networkd.conf configuration from system credentials.
2652 * systemd-networkd will now pick up wireguard secrets from
2655 * systemd-networkd's Varlink API now supports enumerating LLDP peers.
2657 * .link files now support new Property=, ImportProperty=,
2658 UnsetProperty= fields for setting udev properties on a link.
2660 * The various .link files that systemd ships for interfaces that are
2661 supposed to be managed by systemd-networkd only now carry a
2662 ID_NET_MANAGED_BY=io.systemd.Network udev property ensuring that
2663 other network management solutions honouring this udev property do
2664 not come into conflict with networkd, trying to manage these
2667 * .link files now support a new ReceivePacketSteeringCPUMask= setting
2668 for configuring which CPUs to steer incoming packets to.
2670 * The [Network] section in .network files gained a new setting
2671 UseDomains=, which is a single generic knob for controlling the
2672 settings of the same name in the [DHCPv4], [DHCPv6] and
2675 * The 99-default.link file we ship by default (that defines the policy
2676 for all network devices to which no other .link file applies) now
2677 lists "mac" among AlternativeNamesPolicy=. This means that network
2678 interfaces will now by default gain an additional MAC-address based
2679 alternative device name. (i.e. enx…)
2683 * systemd-nspawn now provides a /run/systemd/nspawn/unix-export/
2684 directory where the container payload can expose AF_UNIX sockets to
2685 allow them to be accessed from outside.
2687 * systemd-nspawn will tint the terminal background for containers in a
2688 blueish color. This can be controller with the new --background=
2689 switch or the new $SYSTEMD_TINT_BACKGROUND environment variable.
2691 * systemd-nspawn gained support for the 'owneridmap' option for --bind=
2692 mounts to map the target directory owner from inside the container to
2693 the owner of the directory bound from the host filesystem.
2695 * systemd-nspawn now supports moving Wi-Fi network devices into a
2696 container, just like other network interfaces.
2700 * systemd-resolved now reads RFC 8914 EDE error codes provided by
2701 upstream DNS services.
2703 * systemd-resolved and resolvectl now support RFC 9460 SVCB and HTTPS
2704 records, as well as RFC 2915 NAPTR records.
2706 * resolvectl gained a new option --relax-single-label= to allow
2707 querying single-label hostnames via unicast DNS on a per-query basis.
2709 * systemd-resolved's Varlink IPC interface now supports resolving
2710 DNS-SD services as well as an API for resolving raw DNS RRs.
2712 * systemd-resolved's .dnssd DNS_SD service description files now
2713 support DNS-SD "subtypes" via the new SubType= setting.
2715 * systemd-resolved's configuration may now be reloaded without
2716 restarting the service. (i.e. "systemctl reload systemd-resolved" is
2721 * An sshd_config drop-in to allow ssh keys acquired via userdbctl (for
2722 example expose by homed accounts) to be used for authorization of
2723 incoming SSH connections. This uses the AuthorizedKeysCommand stanza
2724 of sshd_config. Note that sshd only allows a single command to be
2725 configured this way, hence this drop-in might conflict with other
2726 uses of the logic. It is possible to chainload another, similar tool
2727 of another subsystem via the --chain switch of userdbctl, to support
2728 both in parallel. See the "INTEGRATION WITH SSH" section in
2729 userdbctl(1) for details on this. Our recommendation how to combine
2730 other subsystem's use of the SSH authorized keys logic with systemd's
2731 userbctl functionality however is to implement the APIs described
2732 here: https://systemd.io/USER_GROUP_API – in that case this newly
2733 added sshd_config integration would just work and do the right thing
2736 * A small new unit generator "systemd-ssh-generator" has been added. It
2737 checks if the sshd binary is installed. If so, it binds it via
2738 per-connection socket activation to various sockets depending on the
2741 • If the system is run in a VM providing AF_VSOCK support, it
2742 automatically binds sshd to AF_VSOCK port 22.
2744 • If the system is invoked as a full-OS container and the container
2745 manager pre-mounts a directory /run/host/unix-export/, it will
2746 bind sshd to an AF_UNIX socket /run/host/unix-export/ssh. The
2747 idea is the container manager bind mounts the directory to an
2748 appropriate place on the host as well, so that the AF_UNIX socket
2749 may be used to easily connect from the host to the container.
2751 • sshd is also bound to an AF_UNIX socket
2752 /run/ssh-unix-local/socket, which may be to use ssh/sftp in a
2753 "sudo"-like fashion to access resources of other local users.
2755 • Via the kernel command line option "systemd.ssh_listen=" and the
2756 system credential "ssh.listen" sshd may be bound to additional,
2757 explicitly configured options, including AF_INET/AF_INET6 ports.
2759 In particular the first two mechanisms should make dealing with local
2760 VMs and full OS containers a lot easier, as SSH connections will
2761 *just* *work* from the host – even if no networking is available
2764 systemd-ssh-generator optionally generates a per-connection
2765 socket activation service file wrapping sshd. This is only done if
2766 the distribution does not provide one on its own under the name
2767 "sshd@.service". The generated unit only works correctly if the SSH
2768 privilege separation ("privsep") directory exists. Unfortunately
2769 distributions vary wildly where they place this directory. An
2770 incomprehensive list:
2772 • /usr/share/empty.sshd/ (new fedora)
2775 • /run/sshd/ (debian/ubuntu?)
2777 If the SSH privsep directory is placed below /var/ or /run/ care
2778 needs to be taken that the directory is created automatically at boot
2779 if needed, since these directories possibly or always come up
2780 empty. This can be done via a tmpfiles.d/ drop-in. You may use the
2781 "sshdprivsepdir" meson option provided by systemd to configure the
2782 directory, in case you want systemd to create the directory as needed
2783 automatically, if your distribution does not cover this natively.
2785 Recommendations to distributions, in order to make things just work:
2787 • Please provide a per-connection SSH service file under the name
2790 • Please move the SSH privsep dir into /usr/ (so that it is truly
2791 immutable on image-based operating systems, is strictly under
2792 package manager control, and never requires recreation if the
2793 system boots up with an empty /run/ or /var/).
2795 • As an extension of this: please consider following Fedora's lead
2796 here, and use /usr/share/empty.sshd/ to minimize needless
2797 differences between distributions.
2799 • If your distribution insists on placing the directory in /var/ or
2800 /run/ then please at least provide a tmpfiles.d/ drop-in to
2801 recreate it automatically at boot, so that the sshd binary just
2802 works, regardless in which context it is called.
2804 * A small tool "systemd-ssh-proxy" has been added, which is supposed to
2805 act as counterpart to "systemd-ssh-generator". It's a small plug-in
2806 for the SSH client (via ProxyCommand/ProxyUseFdpass) to allow it to
2807 connect to AF_VSOCK or AF_UNIX sockets. Example: "ssh vsock/4711"
2808 connects to a local VM with cid 4711, or "ssh
2809 unix/run/ssh-unix-local/socket" to connect to the local host via the
2810 AF_UNIX socket /run/ssh-unix-local/socket.
2812 systemd-boot and systemd-stub and Related Tools:
2814 * TPM 1.2 PCR measurement support has been removed from systemd-stub.
2815 TPM 1.2 is obsolete and – due to the (by today's standards) weak
2816 cryptographic algorithms it only supports – does not actually provide
2817 the security benefits it's supposed to provide. Given that the rest
2818 of systemd's codebase never supported TPM 1.2, the support has now
2819 been removed from systemd-stub as well.
2821 * systemd-stub will now measure its payload via the new EFI
2822 Confidential Computing APIs (CC), in addition to the pre-existing
2823 measurements to TPM.
2825 * confexts are loaded by systemd-stub from the ESP as well.
2827 * kernel-install gained support for --root= for the 'list' verb.
2829 * bootctl now provides a basic Varlink interface and can be run as a
2830 daemon via a template unit.
2832 * systemd-measure gained new options --certificate=, --private-key=,
2833 and --private-key-source= to allow using OpenSSL's "engines" or
2834 "providers" as the signing mechanism to use when creating signed
2835 TPM2 PCR measurement values.
2837 * ukify gained support for signing of PCR signatures via OpenSSL's
2838 engines and providers.
2840 * ukify now supports zboot kernels.
2842 * systemd-boot now supports passing additional kernel command line
2843 switches to invoked kernels via an SMBIOS Type #11 string
2844 "io.systemd.boot.kernel-cmdline-extra". This is similar to the
2845 pre-existing support for this in systemd-stub, but also applies to
2846 Type #1 Boot Loader Specification Entries.
2848 * systemd-boot's automatic SecureBoot enrollment support gained support
2849 for enrolling "dbx" too (Previously, only db/KEK/PK enrollment was
2850 supported). It also now supports UEFI "Custom" and "Audit" modes.
2852 * The pcrlock policy is saved in an unencrypted credential file
2853 "pcrlock.<entry-token>.cred" under XBOOTLDR/ESP in the
2854 /loader/credentials/ directory. It will be picked up at boot by
2855 systemd-stub and passed to the initrd, where it can be used to unlock
2856 the root file system.
2858 * systemd-pcrlock gained an --entry-token= option to configure the
2861 * systemd-pcrlock now provides a basic Varlink interface and can be run
2862 as a daemon via a template unit.
2864 * systemd-pcrlock's TPM nvindex access policy has been modified, this
2865 means that previous pcrlock policies stored in nvindexes are
2866 invalidated. They must be removed (systemd-pcrlock remove-policy) and
2867 recreated (systemd-pcrlock make-policy). For the time being
2868 systemd-pcrlock remains an experimental feature, but it is expected
2869 to become stable in the next release, i.e. v257.
2871 * systemd-pcrlock's --recovery-pin= switch now takes three values:
2872 "hide", "show", "query". If "show" is selected the automatically
2873 generated recovery PIN is shown to the user. If "query" is selected
2874 then the PIN is queried from the user.
2876 * sd-stub gained support for the new ".ucode" PE section in UKIs, that
2877 may contain CPU microcode data. When control is handed over to the
2878 Linux kernel this data is prepended to the set of initrds passed.
2882 * systemd-run is now a multi-call binary. When invoked as 'run0', it
2883 provides as interface similar to 'sudo', with all arguments starting
2884 at the first non-option parameter being treated the command to invoke
2885 as root. Unlike 'sudo' and similar tools, it does not make use of
2886 setuid binaries or other privilege escalation methods, but instead
2887 runs the specified command as a transient unit, which is started by
2888 the system service manager, so privileges are dropped, rather than
2889 gained, thus implementing a much more robust and safe security
2890 model. As usual, authorization is managed via Polkit.
2892 * systemd-run/run0 will now tint the terminal background on supported
2893 terminals: in a reddish tone when invoking a root service, in a
2894 yellowish tone otherwise. This may be controlled and turned off via
2895 the new --background= switch or the new $SYSTEMD_TINT_BACKGROUND
2896 environment variable.
2898 * systemd-run gained a new option '--ignore-failure' to suppress
2903 * 'systemctl edit --stdin' allows creation of unit files and drop-ins
2904 with contents supplied via standard input. This is useful when creating
2905 configuration programmatically; the tool takes care of figuring out
2906 the file name, creating any directories, and reloading the manager
2909 * 'systemctl disable --now' and 'systemctl mask --now' now work
2910 correctly with template units.
2912 * 'systemd-analyze architectures' lists known CPU architectures.
2914 * 'systemd-analyze --json=…' is supported for 'architectures',
2915 'capability', 'exit-status'.
2917 * 'systemd-tmpfiles --purge' will purge (remove) all files and
2918 directories created via tmpfiles.d configuration.
2920 * systemd-id128 gained new options --no-pager, --no-legend, and
2923 * hostnamectl gained '-j' as shortcut for '--json=pretty' or
2926 * loginctl now supports -j/--json=.
2928 * resolvectl now supports -j/--json= for --type=.
2930 * systemd-tmpfiles gained a new option --dry-run to print what would be
2931 done without actually taking action.
2933 * varlinkctl gained a new --collect switch to collect all responses of
2934 a method call that supports multiple replies and turns it into a
2937 * systemd-dissect gained a new --make-archive option to generate an
2938 archive file (tar.gz and similar) from a disk image.
2942 * systemd-vmspawn gained a new --firmware= option to configure or list
2943 firmware definitions for Qemu, a new --tpm= option to enable or
2944 disable the use of a software TPM, a new --linux= option to specify a
2945 kernel binary for direct kernel boot, a new --initrd= option to
2946 specify an initrd for direct kernel boot, a new -D/--directory option
2947 to use a plain directory as the root file system, a new
2948 --private-users option similar to the one in systemd-nspawn, new
2949 options --bind= and --bind-ro= to bind part of the host's file system
2950 hierarchy into the guest, a new --extra-drive= option to attach
2951 additional storage, and -n/--network-tap/--network-user-mode to
2952 configure networking.
2954 * A new systemd-vmspawn@.service can be used to launch systemd-vmspawn
2957 * systemd-vmspawn gained the new --console= and --background= switches
2958 that control how to interact with the VM. As before, by default an
2959 interactive terminal interface is provided, but now with a background
2960 tinted with a greenish hue.
2962 * systemd-vmspawn can now register its VMs with systemd-machined,
2963 controlled via the --register= switch.
2965 * machinectl's start command (and related) can now invoke images either
2966 as containers via systemd-nspawn (specified as '--runner=nspawn', the
2967 default) or as VMs via systemd-vmspawn (specified as
2968 '--runner=vmspawn' or '-V').
2970 * systemd-vmspawn now supports two switches --pass-ssh-key= and
2971 --ssh-key-type= to optionally set up transient SSH keys to pass to the
2972 invoked VMs in order to be able to SSH into them once booted.
2974 * systemd-vmspawn will now enable various "HyperV enlightenments" and
2975 the "VM Generation ID" on the VMs.
2977 * A new environment variable $SYSTEMD_VMSPAWN_QEMU_EXTRA may carry
2978 additional qemu command line options to pass to qemu.
2980 * systemd-machined gained a new GetMachineSSHInfo() D-Bus method that is
2981 used by systemd-vmspawn to fetch the information needed to ssh into the
2984 * systemd-machined gained a new Varlink interface that is used by
2985 systemd-vmspawn to register machines with additional information and
2990 * systemd-repart gained new options --generate-fstab= and
2991 --generate-crypttab= to write out fstab and crypttab files matching the
2992 generated partitions.
2994 * systemd-repart gained a new option --private-key-source= to allow
2995 using OpenSSL's "engines" or "providers" as the signing mechanism to
2996 use when creating verity signature partitions.
2998 * systemd-repart gained a new DefaultSubvolume= setting in repart.d/
2999 drop-ins that allow configuring the default btrfs subvolume for newly
3000 formatted btrfs file systems.
3004 * libsystemd gained new call sd_bus_creds_new_from_pidfd() to get a
3005 credentials object for a pidfd and sd_bus_creds_get_pidfd_dup() to
3006 retrieve the pidfd from a credentials object.
3008 * sd-bus' credentials logic will now also acquire peer's UNIX group
3009 lists and peer's pidfd if supported and requested.
3011 * RPM macro %_kernel_install_dir has been added with the path
3012 to the directory for kernel-install plugins.
3014 * The liblz4, libzstd, liblzma, libkmod, libgcrypt dependencies have
3015 been changed from regular shared library dependencies into dlopen()
3018 Note that this means that those libraries might not be automatically
3019 pulled in when ELF dependencies are resolved. In particular lack of
3020 libkmod might cause problems with boot. This affects dracut <= 101,
3021 see https://github.com/dracut-ng/dracut-ng/commit/04b362d713235459cf.
3023 * systemd ELF binaries that use libraries via dlopen() are now built with
3024 a new ELF header note section, following a new specification defined at
3025 docs/ELF_DLOPEN_METADATA.md, that provides information about which
3026 sonames are loaded and used if found at runtime. This allows tools and
3027 packagers to programmatically discover the list of optional
3028 dependencies used by all systemd ELF binaries. A parser with packaging
3029 integration tools is available at
3030 https://github.com/systemd/package-notes
3032 * The sd-journal API gained a new call
3033 sd_journal_stream_fd_with_namespace() which is just like
3034 sd_journal_stream_fd() but creates a log stream targeted at a
3035 specific log namespace.
3037 * The sd-id128 API gained a new API call
3038 sd_id128_get_invocation_app_specific() for acquiring an app-specific
3039 ID that is derived from the service invocation ID.
3041 * The sd-event API gained a new API call
3042 sd_event_source_get_inotify_path() that returns the file system path
3043 an inotify event source was created for.
3045 systemd-cryptsetup/systemd-cryptenroll:
3047 * The device node argument to systemd-cryptenroll is now optional. If
3048 omitted it will be derived automatically from the backing block
3049 device of /var/ (which quite likely is the same as the root file
3050 system, hence effectively means if you don't specify things otherwise
3051 the tool will now default to enrolling a key into the root file
3052 system's LUKS device).
3054 * systemd-cryptenroll can now enroll directly with a PKCS11 public key
3055 (instead of a certificate).
3057 * systemd-cryptsetup/systemd-cryptenroll now may lock a disk against a
3058 PKCS#11 provided EC key (before it only supported RSA).
3060 * systemd-cryptsetup gained support for crypttab option
3061 link-volume-key= to link the volume key into the kernel keyring when
3062 the volume is opened.
3064 * systemd-cryptenroll will no longer enable Dictionary Attack
3065 Protection (i.e. turn on NO_DA) for TPM enrollments that do not
3066 involve a PIN. DA should not be necessary in that case (since key
3067 entropy is high enough to make this unnecessary), but risks
3068 accidental lock-out in case of unexpected PCR changes.
3070 * systemd-cryptenroll now supports enrolling a new slot while unlocking
3071 the old slot via TPM2 (previously unlocking only worked via password
3076 * The remaining documentation that was on
3077 https://freedesktop.org/wiki/Software/systemd/ has been moved to
3078 https://systemd.io/.
3080 * A new text describing the VM integration interfaces of systemd has
3083 https://systemd.io/VM_INTERFACE
3085 * The sd_notify() man page has gained examples with C and Python code
3086 that shows how to implement the interface in those languages without
3087 involving libsystemd.
3089 systemd-homed, systemd-logind, systemd-userdbd:
3091 * systemd-homed now supports unlocking of home directories when logging
3092 in via SSH. Previously home directories needed to be unlocked before
3093 an SSH login is attempted.
3095 * JSON User Records have been extended with a separate public storage
3096 area called "User Record Blob Directories". This is intended to store
3097 the user's background image, avatar picture, and other similar items
3098 which are too large to fit into the User Record itself.
3100 systemd-homed, userdbctl, and homectl gained support for blob
3101 directories. homectl gained --avatar= and --login-background= to
3102 control two specific items of the blob directories.
3104 * A new "additionalLanguages" field has been added to JSON user records
3105 (as supported by systemd-homed and systemd-userdbd), which is closely
3106 related to the pre-existing "preferredLanguage", and allows
3107 specifying multiple additional languages for the user account. It is
3108 used to initialize the $LANGUAGES environment variable when used.
3110 * A new pair of "preferredSessionType" and "preferredSessionLauncher"
3111 fields have been added to JSON user records, that may be used to
3112 control which kind of desktop session to preferable activate on
3115 * homectl gained a new verb 'firstboot', and a new
3116 systemd-homed-firstboot.service unit uses this verb to create users
3117 in a first boot environment, either from system credentials or by
3118 querying interactively.
3120 * systemd-logind now supports a new "background-light" session class
3121 which does not pull in the user@.service unit. This is intended in
3122 particular for lighter weight per-user cron jobs which do require any
3123 per-user service manager to be around.
3125 * The per-user service manager will now be tracked as a distinct "manager"
3126 session type among logind sessions of each user.
3128 * homectl now supports an --offline mode, by which certain account
3129 properties can be changed without unlocking the home directory.
3131 * systemd-logind gained a new
3132 org.freedesktop.login1.Manager.ListSessionsEx() method that provides
3133 additional metadata compared to ListSessions(). loginctl makes use of
3134 this to list additional fields in list-sessions.
3136 * systemd-logind gained a new org.freedesktop.login1.Manager.Sleep()
3137 method that automatically redirects to SuspendThenHibernate(),
3138 Suspend(), HybridSleep(), or Hibernate(), depending on what is
3139 supported and configured, a new configuration setting SleepOperation=,
3140 and an accompanying helper method
3141 org.freedesktop.login1.Manager.CanSleep() and property
3142 org.freedesktop.login1.Manager.SleepOperation.
3144 'systemctl sleep' calls the new method to automatically put the
3145 machine to sleep in the most appropriate way.
3147 Credential Management:
3149 * systemd-creds now provides a Varlink IPC API for encrypting and
3150 decrypting credentials.
3152 * systemd-creds' "tpm2-absent" key selection has been renamed to
3153 "null", since that's what it actually does: "encrypt" and "sign"
3154 with a fixed null key. --with-key=null should only be used in very
3155 specific cases, as it provides zero integrity or confidentiality
3156 protections. (i.e. it's only safe to use as fallback in environments
3157 lacking both a TPM and access to the root fs to use the host
3158 encryption key, or when integrity is provided some other way.)
3160 * systemd-creds gained a new switch --allow-null. If specified, the
3161 "decrypt" verb will decode encrypted credentials that use the "null"
3162 key (by default this is refused, since using the "null" key defeats
3163 the authenticated encryption normally done).
3165 Suspend & Hibernate:
3167 * The sleep.conf configuration file gained a new MemorySleepMode=
3168 setting for configuring the sleep mode in more detail.
3170 * A tiny new service systemd-hibernate-clear.service has been added
3171 which clears hibernation information from the HibernateLocation EFI
3172 variable, in case the resume device is gone. Normally, this variable
3173 is supposed to be cleaned up by the code that initiates the resume
3174 from hibernation image. But when the device is missing and that code
3175 doesn't run, this service will now do the necessary work, ensuring
3176 that no outdated hibernation image information remains on subsequent
3179 Unprivileged User Namespaces & Mounts:
3181 * A small new service systemd-nsresourced.service has been added. It
3182 provides a Varlink IPC API that assigns a free, transiently allocated
3183 64K UID/GID range to an uninitialized user namespace a client
3184 provides. It may be used to implement unprivileged container managers
3185 and other programs that need dynamic user ID ranges. It also provides
3186 interfaces to then delegate mount file descriptors, control groups
3187 and network interfaces to user namespaces set up this way.
3189 * A small new service systemd-mountfsd.service has been added. It
3190 provides a Varlink IPC API for mounting DDI images, and returning a set
3191 of mount file descriptors for it. If a user namespace fd is provided
3192 as input, then the mounts are registered with the user namespace. To
3193 ensure trust in the image it must provide Verity information (or
3194 alternatively interactive polkit authentication is required).
3196 * The systemd-dissect tool now can access DDIs fully unprivileged by
3197 using systemd-nsresourced/systemd-mountfsd.
3199 * If the service manager runs unprivileged (i.e. systemd --user) it now
3200 supports RootImage= for accessing DDI images, also implemented via
3201 the systemd-nsresourced/systemd-mountfsd.
3203 * systemd-nspawn may now operate without privileges, if a suitable DDI
3204 is provided via --image=, again implemented via
3205 systemd-nsresourced/systemd-mountfsd.
3209 * timedatectl and machinectl gained option '-P', an alias for
3210 '--value --property=…'.
3212 * Various tools that pretty-print config files will now highlight
3213 configuration directives.
3215 * varlinkctl gained support for the "ssh:" transport. This requires
3216 OpenSSH 9.4 or newer.
3218 * systemd-sysext gained support for enabling system extensions in
3219 mutable fashion, where a writeable upperdir is stored under
3220 /var/lib/extensions.mutable/, and a new --mutable= option to
3221 configure this behaviour. An "ephemeral" mode is not also supported
3222 where the mutable layer is configured to be a tmpfs that is
3223 automatically released when the system extensions are reattached.
3225 * Coredumps are now retained for two weeks by default (instead of three
3228 * portablectl --copy= parameter gained a new 'mixed' argument, that will
3229 result in resources owned by the OS (e.g.: portable profiles) to be linked
3230 but resources owned by the portable image (e.g.: the unit files and the
3231 images themselves) to be copied.
3233 * systemd will now register MIME types for various of its file types
3234 (e.g. journal files, DDIs, encrypted credentials …) via the XDG
3235 shared-mime-info infrastructure. (Files of these types will thus be
3236 recognized as their own thing in desktop file managers such as GNOME
3239 * systemd-dissect will now show the detected sector size of a given DDI
3240 in its default output.
3242 * systemd-portabled now generates recognizable structured log messages
3243 whenever a portable service is attached or detached.
3245 * Verity signature checking in userspace (i.e. checking against
3246 /etc/verity.d/ keys) when activating DDIs can now be turned on/off
3247 via a kernel command line option systemd.allow_userspace_verity= and
3248 an environment variable SYSTEMD_ALLOW_USERSPACE_VERITY=.
3250 * ext4/xfs file system quota handling has been reworked, so that
3251 quotacheck and quotaon are now invoked as per-file-system templated
3252 services (as opposed to single system-wide singletons), similar in
3253 style to the fsck, growfs, pcrfs logic. This means file systems with
3254 quota enabled can now be reasonably enabled at runtime of the system,
3257 * "systemd-analyze dot" will now also show BindsTo= dependencies.
3259 * systemd-debug-generator gained the ability add in arbitrary units
3260 based on them being passed in via system credentials.
3262 * A new kernel command-line option systemd.default_debug_tty= can be
3263 used to specify the TTY for the debug shell, independently of
3264 enabling or disabling it.
3266 * portablectl gained a new --clean switch that clears a portable
3267 service's data (cache, logs, state, runtime, fdstore) when detaching
3270 Contributions from: A S Alam, AKHIL KUMAR,
3271 Abraham Samuel Adekunle, Adrian Vovk, Adrian Wannenmacher,
3272 Alan Liang, Alberto Planas, Alexander Zavyalov, Anders Jonsson,
3273 Andika Triwidada, Andres Beltran, Andrew Sayers,
3274 Antonio Alvarez Feijoo, Arian van Putten, Arthur Zamarin,
3275 Artur Pak, AtariDreams, Benjamin Franzke, Bernhard M. Wiedemann,
3276 Black-Hole1, Bryan Jacobs, Burak Gerz, Carlos Garnacho,
3277 Chandra Pratap, Chris Hofstaedtler, Chris Packham, Chris Simons,
3278 Christian Göttsche, Christian Wesselhoeft, Clayton Craft,
3279 Colin Geniet, Colin Walters, Colin Watson, Costa Tsaousis,
3280 Cristian Rodríguez, Daan De Meyer, Damien Challet, Dan Streetman,
3281 Daniel Winzen, Daniele Medri, David Seifert, David Tardon,
3282 David Venhoek, Diego Viola, Dionna Amalie Glaze,
3283 Dmitry Konishchev, Dmitry V. Levin, Edson Juliano Drosdeck,
3284 Eisuke Kawashima, Eli Schwartz, Emanuele Giuseppe Esposito,
3285 Eric Daigle, Evgeny Vereshchagin, Felix Riemann,
3286 Fernando Fernandez Mancera, Florian Fainelli, Florian Schmaus,
3287 Franck Bui, Frantisek Sumsal, Friedrich Altheide,
3288 Gabríel Arthúr Pétursson, Gaël Donval, Georges Basile Stavracas Neto,
3289 Gerd Hoffmann, GNOME Foundation, Guido Leenders,
3290 Guilhem Lettron, Göran Uddeborg, Hans de Goede, Harald Brinkmann,
3291 Heinrich Schuchardt, Helmut Grohne, Henry Li, Heran Yang,
3292 Holger Assmann, Ivan Kruglov, Ivan Shapovalov, Jakub Sitnicki,
3293 James Muir, Jan Engelhardt, Jan Macku, Jarne Förster, Jeff King,
3294 Jian-Hong Pan, JmbFountain, Joakim Nohlgård, Jonathan Conder,
3295 Julius Alexandre, Jörg Behrmann, Kai Lueke, Kamil Szczęk,
3296 KayJay7, Keian, Kirk, Kristian Klausen, Krzesimir Nowak,
3297 Lain "Fearyncess" Yang, Lars Ellenberg, Lennart Poettering,
3298 Leonard, Luca Boccassi, Lucas Salles, Ludwig Nussel,
3299 Lukáš Nykrýn, Luna Jernberg, Luxiter, Maanya Goenka,
3300 Maciej S. Szmigiero, Mariano Giménez, Markus Merklinger,
3301 Martin Ivicic, Martin Srebotnjak, Martin Trigaux, Martin Wilck,
3302 Mathias Lang, Matt Layher, Matt Muggeridge, Matteo Croce,
3303 Matthias Lisin, Max Gautier, Max Staudt, MaxHearnden,
3304 Michael Biebl, Michal Koutný, Michal Sekletár, Michał Kopeć,
3305 Mike Gilbert, Mike Yuan, Mikko Ylinen, MkfsSion, Moritz Sanft,
3306 MrSmör, Nandakumar Raghavan, Nicholas Little, Nick Cao,
3307 Nick Rosbrook, Nicolas Bouchinet, Norbert Lange,
3308 Ole Peder Brandtzæg, Ondrej Kozina, Oğuz Ersen,
3309 Pablo Méndez Hernández, Pierre GRASSER, Piotr Drąg, QuonXF,
3310 Radoslav Kolev, Rafaël Kooi, Raito Bezarius, Rasmus Villemoes,
3311 Reid Wahl, Renjaya Raga Zenta, Richard Maw, Roland Hieber,
3312 Ronan Pigott, Rose, Ross Burton, Saliba-san, Sam Leonard,
3313 Samuel BF, Sarvajith Adyanthaya, Scrambled 777,
3314 Sebastian Pucilowski, Sergei Zhmylev, Sergey A, Shulhan,
3315 SidhuRupinder, Simon Fowler, Skia, Sludge, Stuart Hayhurst,
3316 Susant Sahani, Takashi Sakamoto, Temuri Doghonadze, Thayne McCombs,
3317 Thilo Fromm, Thomas Blume, Tiago Rocha Cunha, Timo Rothenpieler,
3318 TobiPeterG, Tobias Fleig, Tomáš Pecka, Topi Miettinen,
3319 Tycho Andersen, Unique-Usman, Usman Akinyemi, Vasiliy Kovalev,
3320 Vasiliy Stelmachenok, Victor Berchet, Vishal Chillara Srinivas,
3321 Vitaly Kuznetsov, Vito Caputo, Vladimir Stoiakin, Werner Sembach,
3322 Will Springer, Winterhuman, Xiaotian Wu, Yu Watanabe,
3323 Yuri Chornoivan, Zbigniew Jędrzejewski-Szmek, Zmyeir, anphir,
3324 aslepykh, chenjiayi, cpackham-atlnz, cunshunxia, djantti, drewbug,
3325 hanjinpeng, hfavisado, hulkoba, hydrargyrum, ksaleem, mburucuyapy,
3326 medusalix, mille-feuille, mkubiak, mooo, msizanoen, networkException,
3327 nl6720, r-vdp, runiq, sam-leonard-ct, samuelvw01, sharad3001, spdfnet,
3328 sushmbha, wangyuhang, zeroskyx, zzywysm, İ. Ensar Gülşen,
3329 Łukasz Stelmach, Štěpán Němec, 我超厉害, 김인수
3331 — Edinburgh, 2024-06-11
3335 Announcements of Future Feature Removals and Incompatible Changes:
3337 * Support for split-usr (/usr/ mounted separately during late boot,
3338 instead of being mounted by the initrd before switching to the rootfs)
3339 and unmerged-usr (parallel directories /bin/ and /usr/bin/, /lib/ and
3340 /usr/lib/, …) has been removed. For more details, see:
3341 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
3343 * We intend to remove cgroup v1 support from a systemd release after
3344 the end of 2023. If you run services that make explicit use of
3345 cgroup v1 features (i.e. the "legacy hierarchy" with separate
3346 hierarchies for each controller), please implement compatibility with
3347 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
3348 Most of Linux userspace has been ported over already.
3350 * Support for System V service scripts is now deprecated and will be
3351 removed in a future release. Please make sure to update your software
3352 *now* to include a native systemd unit file instead of a legacy
3353 System V script to retain compatibility with future systemd releases.
3355 * Support for the SystemdOptions EFI variable is deprecated.
3356 'bootctl systemd-efi-options' will emit a warning when used. It seems
3357 that this feature is little-used and it is better to use alternative
3358 approaches like credentials and confexts. The plan is to drop support
3359 altogether at a later point, but this might be revisited based on
3362 * systemd-run's switch --expand-environment= which currently is disabled
3363 by default when combined with --scope, will be changed in a future
3364 release to be enabled by default.
3366 * "systemctl switch-root" is now restricted to initrd transitions only.
3368 Transitions between real systems should be done with
3369 "systemctl soft-reboot" instead.
3371 * The "ip=off" and "ip=none" kernel command line options interpreted by
3372 systemd-network-generator will now result in IPv6RA + link-local
3373 addressing being disabled, too. Previously DHCP was turned off, but
3374 IPv6RA and IPv6 link-local addressing was left enabled.
3376 * The NAMING_BRIDGE_MULTIFUNCTION_SLOT naming scheme has been deprecated
3377 and is now disabled.
3379 * SuspendMode=, HibernateState= and HybridSleepState= in the [Sleep]
3380 section of systemd-sleep.conf are now deprecated and have no effect.
3381 They did not (and could not) take any value other than the respective
3382 default. HybridSleepMode= is also deprecated, and will now always use
3383 the 'suspend' disk mode.
3387 * The way services are spawned has been overhauled. Previously, a
3388 process was forked that shared all of the manager's memory (via
3389 copy-on-write) while doing all the required setup (e.g.: mount
3390 namespaces, CGroup configuration, etc.) before exec'ing the target
3391 executable. This was problematic for various reasons: several glibc
3392 APIs were called that are not supposed to be used after a fork but
3393 before an exec, copy-on-write meant that if either process (the
3394 manager or the child) touched a memory page a copy was triggered, and
3395 also the memory footprint of the child process was that of the
3396 manager, but with the memory limits of the service. From this version
3397 onward, the new process is spawned using CLONE_VM and CLONE_VFORK
3398 semantics via posix_spawn(3), and it immediately execs a new internal
3399 binary, systemd-executor, that receives the configuration to apply
3400 via memfd, and sets up the process before exec'ing the target
3401 executable. The systemd-executor binary is pinned by file descriptor
3402 by each manager instance (system and users), and the reference is
3403 updated on daemon-reexec - it is thus important to reexec all running
3404 manager instances when the systemd-executor and/or libsystemd*
3405 libraries are updated on the filesystem.
3407 * Most of the internal process tracking is being changed to use PIDFDs
3408 instead of PIDs when the kernel supports it, to improve robustness
3411 * A new option SurviveFinalKillSignal= can be used to configure the
3412 unit to be skipped in the final SIGTERM/SIGKILL spree on shutdown.
3413 This is part of the required configuration to let a unit's processes
3414 survive a soft-reboot operation.
3416 * System extension images (sysext) can now set
3417 EXTENSION_RELOAD_MANAGER=1 in their extension-release files to
3418 automatically reload the service manager (PID 1) when
3419 merging/refreshing/unmerging on boot. Generally, while this can be
3420 used to ship services in system extension images it's recommended to
3421 do that via portable services instead.
3423 * The ExtensionImages= and ExtensionDirectories= options now support
3424 confexts images/directories.
3426 * A new option CoredumpReceive= can be set for service and scope units,
3427 together with Delegate=yes, to make systemd-coredump on the host
3428 forward core files from processes crashing inside the delegated
3429 CGroup subtree to systemd-coredump running in the container. This new
3430 option is by default used by systemd-nspawn containers that use the
3433 * A new ConditionSecurity=measured-uki option is now available, to ensure
3434 a unit can only run when the system has been booted from a measured UKI.
3436 * MemoryAvailable= now considers physical memory if there are no CGroup
3437 memory limits set anywhere in the tree.
3439 * The $USER environment variable is now always set for services, while
3440 previously it was only set if User= was specified. A new option
3441 SetLoginEnvironment= is now supported to determine whether to also set
3442 $HOME, $LOGNAME, and $SHELL.
3444 * Socket units now support a new pair of
3445 PollLimitBurst=/PollLimitInterval= options to configure a limit on
3446 how often polling events on the file descriptors backing this unit
3447 will be considered within a time window.
3449 * Scope units can now be created using PIDFDs instead of PIDs to select
3450 the processes they should include.
3452 * Sending SIGRTMIN+18 with 0x500 as sigqueue() value will now cause the
3453 manager to dump the list of currently pending jobs.
3455 * If the kernel supports MOVE_MOUNT_BENEATH, the systemctl and
3456 machinectl bind and mount-image verbs will now cause the new mount to
3457 replace the old mount (if any), instead of overmounting it.
3459 * Units now have MemoryPeak, MemorySwapPeak, MemorySwapCurrent and
3460 MemoryZSwapCurrent properties, which respectively contain the values
3461 of the cgroup v2's memory.peak, memory.swap.peak, memory.swap.current
3462 and memory.zswap.current properties. This information is also shown in
3463 "systemctl status" output, if available.
3465 TPM2 Support + Disk Encryption & Authentication:
3467 * systemd-cryptenroll now allows specifying a PCR bank and explicit hash
3468 value in the --tpm2-pcrs= option.
3470 * systemd-cryptenroll now allows specifying a TPM2 key handle (nv
3471 index) to be used instead of the default SRK via the new
3472 --tpm2-seal-key-handle= option.
3474 * systemd-cryptenroll now allows TPM2 enrollment using only a TPM2
3475 public key (in TPM2B_PUBLIC format) – without access to the TPM2
3476 device itself – which enables offline sealing of LUKS images for a
3477 specific TPM2 chip, as long as the SRK public key is known. Pass the
3478 public to the tool via the new --tpm2-device-key= switch.
3480 * systemd-cryptsetup is now installed in /usr/bin/ and is no longer an
3481 internal-only executable.
3483 * The TPM2 Storage Root Key will now be set up, if not already present,
3484 by a new systemd-tpm2-setup.service early boot service. The SRK will
3485 be stored in PEM format and TPM2_PUBLIC format (the latter is useful
3486 for systemd-cryptenroll --tpm2-device-key=, as mentioned above) for
3487 easier access. A new "srk" verb has been added to systemd-analyze to
3488 allow extracting it on demand if it is already set up.
3490 * The internal systemd-pcrphase executable has been renamed to
3493 * The systemd-pcrextend tool gained a new --pcr= switch to override
3494 which PCR to measure into.
3496 * systemd-pcrextend now exposes a Varlink interface at
3497 io.systemd.PCRExtend that can be used to do measurements and event
3500 * TPM measurements are now also written to an event log at
3501 /run/log/systemd/tpm2-measure.log, using a derivative of the TCG
3502 Canonical Event Log format. Previously we'd only log them to the
3503 journal, where they however were subject to rotation and similar.
3505 * A new component "systemd-pcrlock" has been added that allows managing
3506 local TPM2 PCR policies for PCRs 0-7 and similar, which are hard to
3507 predict by the OS vendor because of the inherently local nature of
3508 what measurements they contain, such as firmware versions of the
3509 system and extension cards and suchlike. pcrlock can predict PCR
3510 measurements ahead of time based on various inputs, such as the local
3511 TPM2 event log, GPT partition tables, PE binaries, UKI kernels, and
3512 various other things. It can then pre-calculate a TPM2 policy from
3513 this, which it stores in an TPM2 NV index. TPM2 objects (such as disk
3514 encryption keys) can be locked against this NV index, so that they
3515 are locked against a specific combination of system firmware and
3516 state. Alternatives for each component are supported to allowlist
3517 multiple kernel versions or boot loader version simultaneously
3518 without losing access to the disk encryption keys. The tool can also
3519 be used to analyze and validate the local TPM2 event log.
3520 systemd-cryptsetup, systemd-cryptenroll, systemd-repart have all been
3521 updated to support such policies. There's currently no support for
3522 locking the system's root disk against a pcrlock policy, this will be
3523 added soon. Moreover, it is currently not possible to combine a
3524 pcrlock policy with a signed PCR policy. This component is
3525 experimental and its public interface is subject to change.
3527 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
3529 * bootctl will now show whether the system was booted from a UKI in its
3532 * systemd-boot and systemd-stub now use different project keys in their
3533 respective SBAT sections, so that they can be revoked individually if
3536 * systemd-boot will no longer load unverified DeviceTree blobs when UEFI
3537 SecureBoot is enabled. For more details see:
3538 https://github.com/systemd/systemd/security/advisories/GHSA-6m6p-rjcq-334c
3540 * systemd-boot gained new hotkeys to reboot and power off the system
3541 from the boot menu ("B" and "O"). If the "auto-poweroff" and
3542 "auto-reboot" options in loader.conf are set these entries are also
3543 shown as menu items (which is useful on devices lacking a regular
3546 * systemd-boot gained a new configuration value "menu-disabled" for the
3547 set-timeout option, to allow completely disabling the boot menu,
3548 including the hotkey.
3550 * systemd-boot will now measure the content of loader.conf in TPM2
3553 * systemd-stub will now concatenate the content of all kernel
3554 command-line addons before measuring them in TPM2 PCR 12, in a single
3555 measurement, instead of measuring them individually.
3557 * systemd-stub will now measure and load DeviceTree Blob addons, which
3558 are searched and loaded following the same model as the existing
3559 kernel command-line addons.
3561 * systemd-stub will now ignore unauthenticated kernel command line options
3562 passed from systemd-boot when running inside Confidential VMs with UEFI
3565 * systemd-stub will now load a DeviceTree blob even if the firmware did
3566 not load any beforehand (e.g.: for ACPI systems).
3568 * ukify is no longer considered experimental, and now ships in /usr/bin/.
3570 * ukify gained a new verb inspect to describe the sections of a UKI and
3571 print the contents of the well-known sections.
3573 * ukify gained a new verb genkey to generate a set of key pairs for
3574 signing UKIs and their PCR data.
3576 * The 90-loaderentry kernel-install hook now supports installing device
3579 * kernel-install now supports the --json=, --root=, --image=, and
3580 --image-policy= options for the inspect verb.
3582 * kernel-install now supports new list and add-all verbs. The former
3583 lists all installed kernel images (if those are available in
3584 /usr/lib/modules/). The latter will install all the kernels it can
3589 * A new option --copy-from= has been added that synthesizes partition
3590 definitions from the given image, which are then applied by the
3591 systemd-repart algorithm.
3593 * A new option --copy-source= has been added, which can be used to specify
3594 a directory to which CopyFiles= is considered relative to.
3596 * New --make-ddi=confext, --make-ddi=sysext, and --make-ddi=portable
3597 options have been added to make it easier to generate these types of
3598 DDIs, without having to provide repart.d definitions for them.
3600 * The dm-verity salt and UUID will now be derived from the specified
3603 * New VerityDataBlockSizeBytes= and VerityHashBlockSizeBytes= can now be
3604 configured in repart.d/ configuration files.
3606 * A new Subvolumes= setting is now supported in repart.d/ configuration
3607 files, to indicate which directories in the target partition should be
3610 * A new --tpm2-device-key= option can be used to lock a disk against a
3611 specific TPM2 public key. This matches the same switch the
3612 systemd-cryptenroll tool now supports (see above).
3616 * The journalctl --lines= parameter now accepts +N to show the oldest N
3617 entries instead of the newest.
3619 * journald now ensures that sealing happens once per epoch, and sets a
3620 new compatibility flag to distinguish old journal files that were
3621 created before this change, for backward compatibility.
3625 * udev will now create symlinks to loopback block devices in the
3626 /dev/disk/by-loop-ref/ directory that are based on the .lo_file_name
3627 string field selected during allocation. The systemd-dissect tool and
3628 the util-linux losetup command now supports a complementing new switch
3629 --loop-ref= for selecting the string. This means a loopback block
3630 device may now be allocated under a caller-chosen reference and can
3631 subsequently be referenced without first having to look up the block
3632 device name the caller ended up with.
3634 * udev also creates symlinks to loopback block devices in the
3635 /dev/disk/by-loop-inode/ directory based on the .st_dev/st_ino fields
3636 of the inode attached to the loopback block device. This means that
3637 attaching a file to a loopback device will implicitly make a handle
3638 available to be found via that file's inode information.
3640 * udevadm info gained support for JSON output via a new --json= flag, and
3641 for filtering output using the same mechanism that udevadm trigger
3644 * The predictable network interface naming logic is extended to include
3645 the SR-IOV-R "representor" information in network interface names.
3646 This feature was intended for v254, but even though the code was
3647 merged, the part that actually enabled the feature was forgotten.
3648 It is now enabled by default and is part of the new "v255" naming
3651 * A new hwdb/rules file has been added that sets the
3652 ID_NET_AUTO_LINK_LOCAL_ONLY=1 udev property on all network interfaces
3653 that should usually only be configured with link-local addressing
3654 (IPv4LL + IPv6LL), i.e. for PC-to-PC cables ("laplink") or
3655 Thunderbolt networking. systemd-networkd and NetworkManager (soon)
3656 will make use of this information to apply an appropriate network
3657 configuration by default.
3659 * The ID_NET_DRIVER property on network interfaces is now set
3660 relatively early in the udev rule set so that other rules may rely on
3661 its use. This is implemented in a new "net-driver" udev built-in.
3665 * The "duid-only" option for DHCPv4 client's ClientIdentifier= setting
3666 is now dropped, as it never worked, hence it should not be used by
3669 * The 'prefixstable' ipv6 address generation mode now considers the SSID
3670 when generating stable addresses, so that a different stable address
3671 is used when roaming between wireless networks. If you already use
3672 'prefixstable' addresses with wireless networks, the stable address
3673 will be changed by the update.
3675 * The DHCPv4 client gained a RapidCommit option, true by default, which
3676 enables RFC4039 Rapid Commit behavior to obtain a lease in a
3677 simplified 2-message exchange instead of the typical 4-message
3678 exchange, if also supported by the DHCP server.
3680 * The DHCPv4 client gained new InitialCongestionWindow= and
3681 InitialAdvertisedReceiveWindow= options for route configurations.
3683 * The DHCPv4 client gained a new RequestAddress= option that allows
3684 to send a preferred IP address in the initial DHCPDISCOVER message.
3686 * The DHCPv4 server and client gained support for IPv6-only mode
3689 * The SendHostname= and Hostname= options are now available for the
3690 DHCPv6 client, independently of the DHCPv4= option, so that these
3691 configuration values can be set independently for each client.
3693 * The DHCPv4 and DHCPv6 client state can now be queried via D-Bus,
3694 including lease information.
3696 * The DHCPv6 client can now be configured to use a custom DUID type.
3698 * .network files gained a new IPv4ReversePathFilter= setting in the
3699 [Network] section, to control sysctl's rp_filter setting.
3701 * .network files gaiend a new HopLimit= setting in the [Route] section,
3702 to configure a per-route hop limit.
3704 * .network files gained a new TCPRetransmissionTimeoutSec= setting in
3705 the [Route] section, to configure a per-route TCP retransmission
3708 * A new directive NFTSet= provides a method for integrating network
3709 configuration into firewall rules with NFT sets. The benefit of using
3710 this setting is that static network configuration or dynamically
3711 obtained network addresses can be used in firewall rules with the
3712 indirection of NFT set types.
3714 * The [IPv6AcceptRA] section supports the following new options:
3715 UsePREF64=, UseHopLimit=, UseICMP6RateLimit=, and NFTSet=.
3717 * The [IPv6SendRA] section supports the following new options:
3718 RetransmitSec=, HopLimit=, HomeAgent=, HomeAgentLifetimeSec=, and
3719 HomeAgentPreference=.
3721 * A new [IPv6PREF64Prefix] set of options, containing Prefix= and
3722 LifetimeSec=, has been introduced to append pref64 options in router
3723 advertisements (RFC8781).
3725 * The network generator now configures the interfaces with only
3726 link-local addressing if "ip=link-local" is specified on the kernel
3729 * The prefix of the configuration files generated by the network
3730 generator from the kernel command line is now prefixed with '70-',
3731 to make them have higher precedence over the default configuration
3734 * Added a new -Ddefault-network=BOOL meson option, that causes more
3735 .network files to be installed as enabled by default. These configuration
3736 files will which match generic setups, e.g. 89-ethernet.network matches
3737 all Ethernet interfaces and enables both DHCPv4 and DHCPv6 clients.
3739 * If a ID_NET_MANAGED_BY= udev property is set on a network device and
3740 it is any other string than "io.systemd.Network" then networkd will
3741 not manage this device. This may be used to allow multiple network
3742 management services to run in parallel and assign ownership of
3743 specific devices explicitly. NetworkManager will soon implement a
3746 * .network files gained a new MulticastIGMPVersion= setting in the
3747 [Network] section, to control sysctl's
3748 /proc/sys/net/ipv4/conf/INTERFACE/force_igmp_version setting.
3752 * systemctl is-failed now checks the system state if no unit is
3755 * systemctl will now automatically soft-reboot if a new root file system
3756 is found under /run/nextroot/ when a reboot operation is invoked.
3760 * Wall messages now work even when utmp support is disabled, using
3761 systemd-logind to query the necessary information.
3763 * systemd-logind now sends a new PrepareForShutdownWithMetadata D-Bus
3764 signal before shutdown/reboot/soft-reboot that includes additional
3765 information compared to the PrepareForShutdown signal. Currently the
3766 additional information is the type of operation that is about to be
3769 Hibernation & Suspend:
3771 * The kernel and OS versions will no longer be checked on resume from
3774 * Hibernation into swap files backed by btrfs is now supported.
3775 (Previously this was supported only for other file systems.)
3779 * A new systemd-vmspawn tool has been added, that aims to provide for VMs
3780 the same interfaces and functionality that systemd-nspawn provides for
3781 containers. For now it supports QEMU as a backend, and exposes some of
3782 its options to the user. This component is experimental and its public
3783 interface is subject to change.
3785 * "systemd-analyze plot" has gained tooltips on each unit name with
3786 related-unit information in its svg output, such as Before=,
3787 Requires=, and similar properties.
3789 * A new varlinkctl tool has been added to allow interfacing with
3790 Varlink services, and introspection has been added to all such
3791 services. This component is experimental and its public interface is
3794 * systemd-sysext and systemd-confext now expose a Varlink service
3795 at io.systemd.sysext.
3797 * portable services now accept confexts as extensions.
3799 * systemd-sysupdate now accepts directories in the MatchPattern= option.
3801 * systemd-run will now output the invocation ID of the launched
3802 transient unit and its peak memory usage.
3804 * systemd-analyze, systemd-tmpfiles, systemd-sysusers, systemd-sysctl,
3805 and systemd-binfmt gained a new --tldr option that can be used instead
3806 of --cat-config to suppress uninteresting configuration lines, such as
3807 comments and whitespace.
3809 * resolvectl gained a new "show-server-state" command that shows
3810 current statistics of the resolver. This is backed by a new
3811 DumpStatistics() Varlink method provided by systemd-resolved.
3813 * systemd-timesyncd will now emit a D-Bus signal when the LinkNTPServers
3816 * vconsole now supports KEYMAP=@kernel for preserving the kernel keymap
3819 * seccomp now supports the LoongArch64 architecture.
3821 * seccomp may now be enabled for services running as a non-root User=
3822 without NoNewPrivileges=yes.
3824 * systemd-id128 now supports a new -P option to show only values. The
3825 combination of -P and --app options is also supported.
3827 * A new pam_systemd_loadkey.so PAM module is now available, which will
3828 automatically fetch the passphrase used by cryptsetup to unlock the
3829 root file system and set it as the PAM authtok. This enables, among
3830 other things, configuring auto-unlock of the GNOME Keyring / KDE
3831 Wallet when autologin is configured.
3833 * Many meson options now use the 'feature' type, which means they
3834 take enabled/disabled/auto as values.
3836 * A new meson option -Dconfigfiledir= can be used to change where
3837 configuration files with default values are installed to.
3839 * Options and verbs in man pages are now tagged with the version they
3840 were first introduced in.
3842 * A new component "systemd-storagetm" has been added, which exposes all
3843 local block devices as NVMe-TCP devices, fully automatically. It's
3844 hooked into a new target unit storage-target-mode.target that is
3845 suppsoed to be booted into via
3846 rd.systemd.unit=storage-target-mode.target on the kernel command
3847 line. This is intended to be used for installers and debugging to
3848 quickly get access to the local disk. It's inspired by MacOS "target
3849 disk mode". This component is experimental and its public interface is
3852 * A new component "systemd-bsod" has been added, which can show logged
3853 error messages full screen, if they have a log level of LOG_EMERG log
3854 level. This component is experimental and its public interface is
3857 * The systemd-dissect tool's --with command will now set the
3858 $SYSTEMD_DISSECT_DEVICE environment variable to the block device it
3859 operates on for the invoked process.
3861 * The systemd-mount tool gained a new --tmpfs switch for mounting a new
3862 'tmpfs' instance. This is useful since it does so via .mount units
3863 and thus can be executed remotely or in containers.
3865 * The various tools in systemd that take "verbs" (such as systemctl,
3866 loginctl, machinectl, …) now will suggest a close verb name in case
3867 the user specified an unrecognized one.
3869 * libsystemd now exports a new function sd_id128_get_app_specific()
3870 that generates "app-specific" 128bit IDs from any ID. It's similar to
3871 sd_id128_get_machine_app_specific() and
3872 sd_id128_get_boot_app_specific() but takes the ID to base calculation
3873 on as input. This new functionality is also exposed in the
3874 systemd-id128 tool where you can now combine --app= with 'show'.
3876 * All tools that parse timestamps now can also parse RFC3339 style
3877 timestamps that include the "T" and Z" characters.
3879 * New documentation has been added:
3881 https://systemd.io/FILE_DESCRIPTOR_STORE
3882 https://systemd.io/TPM2_PCR_MEASUREMENTS
3883 https://systemd.io/MOUNT_REQUIREMENTS
3885 * The codebase now recognizes the suffix .confext.raw and .sysext.raw
3886 as alternative to the .raw suffix generally accepted for DDIs. It is
3887 recommended to name configuration extensions and system extensions
3888 with such suffixes, to indicate their purpose in the name.
3890 * The sd-device API gained a new function
3891 sd_device_enumerator_add_match_property_required() which allows
3892 configuring matches on properties that are strictly required. This is
3893 different from the existing sd_device_enumerator_add_match_property()
3894 matches of which one needs to apply.
3896 * The MAC address the veth side of an nspawn container shall get
3897 assigned may now be controlled via the $SYSTEMD_NSPAWN_NETWORK_MAC
3898 environment variable.
3900 * The libiptc dependency is now implemented via dlopen(), so that tools
3901 such as networkd and nspawn no longer have a hard dependency on the
3902 shared library when compiled with support for libiptc.
3904 * New rpm macros have been added: %systemd_user_daemon_reexec does
3905 daemon-reexec for all user managers, and %systemd_postun_with_reload
3906 and %systemd_user_postun_with_reload do a reload for system and user
3909 * coredumpctl now propagates SIGTERM to the debugger process.
3911 Contributions from: 김인수, Abderrahim Kitouni, Adam Goldman,
3912 Adam Williamson, Alexandre Peixoto Ferreira, Alex Hudspith,
3913 Alvin Alvarado, André Paiusco, Antonio Alvarez Feijoo,
3914 Anton Lundin, Arian van Putten, Arseny Maslennikov, Arthur Shau,
3915 Balázs Úr, beh_10257, Benjamin Peterson, Bertrand Jacquin,
3916 Brian Norris, Charles Lee, Cheng-Chia Tseng, Chris Patterson,
3917 Christian Hergert, Christian Hesse, Christian Kirbach,
3918 Clayton Craft, commondservice, cunshunxia, Curtis Klein, cvlc12,
3919 Daan De Meyer, Daniele Medri, Daniel P. Berrangé, Daniel Rusek,
3920 Daniel Thompson, Dan Nicholson, Dan Streetman, David Rheinsberg,
3921 David Santamaría Rogado, David Tardon, dependabot[bot],
3922 Diego Viola, Dmitry V. Levin, Emanuele Giuseppe Esposito,
3923 Emil Renner Berthing, Emil Velikov, Etienne Dechamps, Fabian Vogt,
3924 felixdoerre, Felix Dörre, Florian Schmaus, Franck Bui,
3925 Frantisek Sumsal, G2-Games, Gioele Barabucci, Hugo Carvalho,
3926 huyubiao, Iago López Galeiras, IllusionMan1212, Jade Lovelace,
3927 janana, Jan Janssen, Jan Kuparinen, Jan Macku, Jeremy Fleischman,
3928 Jin Liu, jjimbo137, Joerg Behrmann, Johannes Segitz, Jordan Rome,
3929 Jordan Williams, Julien Malka, Juno Computers, Khem Raj, khm,
3930 Kingbom Dou, Kiran Vemula, Krzesimir Nowak, Laszlo Gombos,
3931 Lennart Poettering, linuxlion, Luca Boccassi, Lucas Adriano Salles,
3932 Lukas, Lukáš Nykrýn, Maanya Goenka, Maarten, Malte Poll,
3933 Marc Pervaz Boocha, Martin Beneš, Martin Joerg, Martin Wilck,
3934 Mathieu Tortuyaux, Matthias Schiffer, Maxim Mikityanskiy,
3935 Max Kellermann, Michael A Cassaniti, Michael Biebl, Michael Kuhn,
3936 Michael Vasseur, Michal Koutný, Michal Sekletár, Mike Yuan,
3937 Milton D. Miller II, mordner, msizanoen, NAHO, Nandakumar Raghavan,
3938 Neil Wilson, Nick Rosbrook, Nils K, NRK, Oğuz Ersen,
3939 Omojola Joshua, onenowy, Paul Meyer, Paymon MARANDI, pelaufer,
3940 Peter Hutterer, PhylLu, Pierre GRASSER, Piotr Drąg, Priit Laes,
3941 Rahil Bhimjiani, Raito Bezarius, Raul Cheleguini, Reto Schneider,
3942 Richard Maw, Robby Red, RoepLuke, Roland Hieber, Roland Singer,
3943 Ronan Pigott, Sam James, Sam Leonard, Sergey A, Susant Sahani,
3944 Sven Joachim, Tad Fisher, Takashi Sakamoto, Thorsten Kukuk, Tj,
3945 Tomasz Świątek, Topi Miettinen, Valentin David,
3946 Valentin Lefebvre, Victor Westerhuis, Vincent Haupert,
3947 Vishal Chillara Srinivas, Vito Caputo, Warren, Weblate,
3948 Xiaotian Wu, xinpeng wang, Yaron Shahrabani, Yo-Jung Lin,
3949 Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zeroskyx,
3950 Дамјан Георгиевски, наб
3952 — Edinburgh, 2023-12-06
3956 Announcements of Future Feature Removals and Incompatible Changes:
3958 * The next release (v255) will remove support for split-usr (/usr/
3959 mounted separately during late boot, instead of being mounted by the
3960 initrd before switching to the rootfs) and unmerged-usr (parallel
3961 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
3963 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
3965 * We intend to remove cgroup v1 support from a systemd release after
3966 the end of 2023. If you run services that make explicit use of
3967 cgroup v1 features (i.e. the "legacy hierarchy" with separate
3968 hierarchies for each controller), please implement compatibility with
3969 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
3970 Most of Linux userspace has been ported over already.
3972 * Support for System V service scripts is now deprecated and will be
3973 removed in a future release. Please make sure to update your software
3974 *now* to include a native systemd unit file instead of a legacy
3975 System V script to retain compatibility with future systemd releases.
3977 * Support for the SystemdOptions EFI variable is deprecated.
3978 'bootctl systemd-efi-options' will emit a warning when used. It seems
3979 that this feature is little-used and it is better to use alternative
3980 approaches like credentials and confexts. The plan is to drop support
3981 altogether at a later point, but this might be revisited based on
3984 * EnvironmentFile= now treats the line following a comment line
3985 trailing with escape as a non comment line. For details, see:
3986 https://github.com/systemd/systemd/issues/27975
3988 * PrivateNetwork=yes and NetworkNamespacePath= now imply
3989 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
3991 * Behaviour of sandboxing options for the per-user service manager
3992 units has changed. They now imply PrivateUsers=yes, which means user
3993 namespaces will be implicitly enabled when a sandboxing option is
3994 enabled in a user unit. Enabling user namespaces has the drawback
3995 that system users will no longer be visible (and processes/files will
3996 appear as owned by 'nobody') in the user unit.
3998 By definition a sandboxed user unit should run with reduced
3999 privileges, so impact should be small. This will remove a great
4000 source of confusion that has been reported by users over the years,
4001 due to how these options require an extra setting to be manually
4002 enabled when used in the per-user service manager, which is not
4003 needed in the system service manager. For more details, see:
4004 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
4006 * systemd-run's switch --expand-environment= which currently is disabled
4007 by default when combined with --scope, will be changed in a future
4008 release to be enabled by default.
4010 Security Relevant Changes:
4012 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
4013 process capability to invoked session processes of regular users on
4014 local seats (as well as to systemd --user), unless configured
4015 otherwise via data from JSON user records, or via the PAM module's
4016 parameter list. This is useful in order allow desktop tools such as
4017 GNOME's Alarm Clock application to set a timer for
4018 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
4019 per-user service unit file may thus use AmbientCapability= to pass
4020 the capability to invoked processes. Note that this capability is
4021 relatively narrow in focus (in particular compared to other process
4022 capabilities such as CAP_SYS_ADMIN) and we already — by default —
4023 permit more impactful operations such as system suspend to local
4028 * Memory limits that apply while the unit is activating are now
4029 supported. Previously IO and CPU settings were already supported via
4030 StartupCPUWeight= and similar. The same logic has been added for the
4031 various manager and unit memory settings (DefaultStartupMemoryLow=,
4032 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
4033 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
4035 * The service manager gained support for enqueuing POSIX signals to
4036 services that carry an additional integer value, exposing the
4037 sigqueue() system call. This is accessible via new D-Bus calls
4038 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
4039 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
4040 via the new --kill-value= option.
4042 * systemctl gained a new "list-paths" verb, which shows all currently
4043 active .path units, similarly to how "systemctl list-timers" shows
4044 active timers, and "systemctl list-sockets" shows active sockets.
4046 * systemctl gained a new --when= switch which is honoured by the various
4047 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
4048 scheduling these operations by time, similar in fashion to how this
4049 has been supported by SysV shutdown.
4051 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
4052 supports the new PR_SET_MDWE prctl() call, it is used instead of the
4053 seccomp()-based system call filter to achieve the same effect.
4055 * A new set of kernel command line options is now understood:
4056 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
4057 systemd.tty.columns.<name>= allow configuring the TTY type and
4058 dimensions for the tty specified via <name>. When systemd invokes a
4059 service on a tty (via TTYName=) it will look for these and configure
4060 the TTY accordingly. This is particularly useful in VM environments
4061 to propagate host terminal settings into the appropriate TTYs of the
4064 * A new RootEphemeral= setting is now understood in service units. It
4065 takes a boolean argument. If enabled for services that use RootImage=
4066 or RootDirectory= an ephemeral copy of the disk image or directory
4067 tree is made when the service is started. It is removed automatically
4068 when the service is stopped. That ephemeral copy is made using
4069 btrfs/xfs reflinks or btrfs snapshots, if available.
4071 * The service activation logic gained new settings RestartSteps= and
4072 RestartMaxDelaySec= which allow exponentially-growing restart
4073 intervals for Restart=.
4075 * The service activation logic gained a new setting RestartMode= which
4076 can be set to 'direct' to skip the inactive/failed states when
4077 restarting, so that dependent units are not notified until the service
4078 converges to a final (successful or failed) state. For example, this
4079 means that OnSuccess=/OnFailure= units will not be triggered until the
4080 service state has converged.
4082 * PID 1 will now automatically load the virtio_console kernel module
4083 during early initialization if running in a suitable VM. This is done
4084 so that early-boot logging can be written to the console if available.
4086 * Similarly, virtio-vsock support is loaded early in suitable VM
4087 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
4088 to the VMM if configured, thus loading this early is beneficial.
4090 * A new verb "fdstore" has been added to systemd-analyze to show the
4091 current contents of the file descriptor store of a unit. This is
4092 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
4093 the service manager.
4095 * The service manager will now set a new $FDSTORE environment variable
4096 when invoking processes for services that have the file descriptor
4099 * A new service option FileDescriptorStorePreserve= has been added that
4100 allows tuning the lifecycle of the per-service file descriptor store.
4101 If set to "yes", the entries in the fd store are retained even after
4102 the service has been fully stopped.
4104 * The "systemctl clean" command may now be used to clear the fdstore of
4107 * Unit *.preset files gained a new directive "ignore", in addition to
4108 the existing "enable" and "disable". As the name suggests, matching
4109 units are left unchanged, i.e. neither enabled nor disabled.
4111 * Service units gained a new setting DelegateSubgroup=. It takes the
4112 name of a sub-cgroup to place any processes the service manager forks
4113 off in. Previously, the service manager would place all service
4114 processes directly in the top-level cgroup it created for the
4115 service. This usually meant that main process in a service with
4116 delegation enabled would first have to create a subgroup and move
4117 itself down into it, in order to not conflict with the "no processes
4118 in inner cgroups" rule of cgroup v2. With this option, this step is
4119 now handled by PID 1.
4121 * The service manager will now look for .upholds/ directories,
4122 similarly to the existing support for .wants/ and .requires/
4123 directories. Symlinks in this directory result in Upholds=
4126 The [Install] section of unit files gained support for a new
4127 UpheldBy= directive to generate .upholds/ symlinks automatically when
4130 * The service manager now supports a new kernel command line option
4131 systemd.default_device_timeout_sec=, which may be used to override
4132 the default timeout for .device units.
4134 * A new "soft-reboot" mechanism has been added to the service manager.
4135 A "soft reboot" is similar to a regular reboot, except that it
4136 affects userspace only: the service manager shuts down any running
4137 services and other units, then optionally switches into a new root
4138 file system (mounted to /run/nextroot/), and then passes control to a
4139 systemd instance in the new file system which then starts the system
4140 up again. The kernel is not rebooted and neither is the hardware,
4141 firmware or boot loader. This provides a fast, lightweight mechanism
4142 to quickly reset or update userspace, without the latency that a full
4143 system reset involves. Moreover, open file descriptors may be passed
4144 across the soft reboot into the new system where they will be passed
4145 back to the originating services. This allows pinning resources
4146 across the reboot, thus minimizing grey-out time further. This new
4147 reboot mechanism is accessible via the new "systemctl soft-reboot"
4150 * Services using RootDirectory= or RootImage= will now have read-only
4151 access to a copy of the host's os-release file under
4152 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
4153 This was already the case for Portable Services, and the feature has
4154 now been extended to all services that do not run off the host's
4157 * A new service setting MemoryKSM= has been added to enable kernel
4158 same-page merging individually for services.
4160 * A new service setting ImportCredentials= has been added that augments
4161 LoadCredential= and LoadCredentialEncrypted= and searches for
4162 credentials to import from the system, and supports globbing.
4164 * A new job mode "restart-dependencies" has been added to the service
4165 manager (exposed via systemctl --job-mode=). It is only valid when
4166 used with "start" jobs, and has the effect that the "start" job will
4167 be propagated as "restart" jobs to currently running units that have
4168 a BindsTo= or Requires= dependency on the started unit.
4170 * A new verb "whoami" has been added to "systemctl" which determines as
4171 part of which unit the command is being invoked. It writes the unit
4172 name to standard output. If one or more PIDs are specified reports
4173 the unit names the processes referenced by the PIDs belong to.
4175 * The system and service credential logic has been improved: there's
4176 now a clearly defined place where system provisioning tools running
4177 in the initrd can place credentials that will be imported into the
4178 system's set of credentials during the initrd → host transition: the
4179 /run/credentials/@initrd/ directory. Once the credentials placed
4180 there are imported into the system credential set they are deleted
4181 from this directory, and the directory itself is deleted afterwards
4184 * A new kernel command line option systemd.set_credential_binary= has
4185 been added, that is similar to the pre-existing
4186 systemd.set_credential= but accepts arbitrary binary credential data,
4187 encoded in Base64. Note that the kernel command line is not a
4188 recommend way to transfer credentials into a system, since it is
4189 world-readable from userspace.
4191 * The default machine ID to use may now be configured via the
4192 system.machine_id system credential. It will only be used if no
4193 machine ID was set yet on the host.
4195 * On Linux kernel 6.4 and newer system and service credentials will now
4196 be placed in a tmpfs instance that has the "noswap" mount option
4197 set. Previously, a "ramfs" instance was used. By switching to tmpfs
4198 ACL support and overall size limits can now be enforced, without
4199 compromising on security, as the memory is never paged out either
4202 * The service manager now can detect when it is running in a
4203 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
4204 accepted by ConditionSecurity= for units that want to conditionalize
4205 themselves on this. systemd-detect-virt gained new 'cvm' and
4206 '--list-cvm' switches to respectively perform the detection or list
4207 all known flavours of confidential VM, depending on the vendor. The
4208 manager will publish a 'ConfidentialVirtualization' D-Bus property,
4209 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
4210 variable for unit generators. Finally, udev rules can match on a new
4211 'cvm' key that will be set when in a confidential VM.
4212 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
4213 strings and QEMU's fw_cfg protocol will not be used to import
4214 credentials and kernel command line parameters by the system manager,
4215 systemd-boot and systemd-stub, because the hypervisor is considered
4216 untrusted in this particular setting.
4220 * The sd-journal API gained a new call sd_journal_get_seqnum() to
4221 retrieve the current log record's sequence number and sequence number
4222 ID, which allows applications to order records the same way as
4223 journal does internally. The sequence number is now also exported in
4224 the JSON and "export" output of the journal.
4226 * journalctl gained a new switch --truncate-newline. If specified
4227 multi-line log records will be truncated at the first newline,
4228 i.e. only the first line of each log message will be shown.
4230 * systemd-journal-upload gained support for --namespace=, similar to
4231 the switch of the same name of journalctl.
4235 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
4236 may be used to exclude certain files from the effect of CopyFiles=.
4238 * systemd-repart's Verity support now implements the Minimize= setting
4239 to minimize the size of the resulting partition.
4241 * systemd-repart gained a new --offline= switch, which may be used to
4242 control whether images shall be built "online" or "offline",
4243 i.e. whether to make use of kernel facilities such as loopback block
4244 devices and device mapper or not.
4246 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
4247 partition with some files, it will now default to VFAT rather than
4250 * systemd-repart gained a new --architecture= switch. If specified, the
4251 per-architecture GPT partition types (i.e. the root and /usr/
4252 partitions) configured in the partition drop-in files are
4253 automatically adjusted to match the specified CPU architecture, in
4254 order to simplify cross-architecture DDI building.
4256 * systemd-repart will now default to a minimum size of 300MB for XFS
4257 filesystems if no size parameter is specified. This matches what the
4258 XFS tools (xfsprogs) can support.
4260 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
4262 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
4263 Instead, pyelftools is now needed, and it will be used to perform the
4264 ELF -> PE relocations at build time.
4266 * bootctl gained a new switch --print-root-device/-R that prints the
4267 block device the root file system is backed by. If specified twice,
4268 it returns the whole disk block device (as opposed to partition block
4269 device) the root file system is on. It's useful for invocations such
4270 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
4273 * systemd-stub will now look for the SMBIOS Type 1 field
4274 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
4275 kernel command line it invokes. This is useful for VMMs such as qemu
4276 to pass additional kernel command lines into the system even when
4277 booting via full UEFI. The contents of the field are measured into
4280 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
4281 value "auto". With this value, a kernel will be automatically
4282 analyzed, and if it qualifies as UKI, it will be installed as if the
4283 setting was to set to "uki", otherwise as "bls".
4285 * systemd-stub can now optionally load UEFI PE "add-on" images that may
4286 contain additional kernel command line information. These "add-ons"
4287 superficially look like a regular UEFI executable, and are expected
4288 to be signed via SecureBoot/shim. However, they do not actually
4289 contain code, but instead a subset of the PE sections that UKIs
4290 support. They are supposed to provide a way to extend UKIs with
4291 additional resources in a secure and authenticated way. Currently,
4292 only the .cmdline PE section may be used in add-ons, in which case
4293 any specified string is appended to the command line embedded into
4294 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
4295 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
4296 the future we expect other sections to be made extensible like this as
4299 * ukify has been updated to allow building these UEFI PE "add-on"
4300 images, using the new 'addon<EFI-ARCH>.efi.stub'.
4302 * ukify now accepts SBAT information to place in the .sbat PE section
4303 of UKIs and addons. If a UKI is built the SBAT information from the
4304 inner kernel is merged with any SBAT information associated with
4305 systemd-stub and the SBAT data specified on the ukify command line.
4307 * The kernel-install script has been rewritten in C, and reuses much of
4308 the infrastructure of existing tools such as bootctl. It also gained
4309 --esp-path= and --boot-path= options to override the path to the ESP,
4310 and the $BOOT partition. Options --make-entry-directory= and
4311 --entry-token= have been added as well, similar to bootctl's options
4314 * A new kernel-install plugin 60-ukify has been added which will
4315 combine kernel/initrd locally into a UKI and optionally sign them
4316 with a local key. This may be used to switch to UKI mode even on
4317 systems where a local kernel or initrd is used. (Typically UKIs are
4318 built and signed by the vendor.)
4320 * The ukify tool now supports "pesign" in addition to the pre-existing
4321 "sbsign" for signing UKIs.
4323 * systemd-measure and systemd-stub now look for the .uname PE section
4324 that should contain the kernel's "uname -r" string.
4326 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
4327 "offline", i.e. without access to a TPM (physical or
4330 Memory Pressure & Control:
4332 * The sd-event API gained new calls sd_event_add_memory_pressure(),
4333 sd_event_source_set_memory_pressure_type(),
4334 sd_event_source_set_memory_pressure_period() to create and configure
4335 an event source that is called whenever the OS signals memory
4336 pressure. Another call sd_event_trim_memory() is provided that
4337 compacts the process' memory use by releasing allocated but unused
4338 malloc() memory back to the kernel. Services can also provide their
4339 own custom callback to do memory trimming. This should improve system
4340 behaviour under memory pressure, as on Linux traditionally provided
4341 no mechanism to return process memory back to the kernel if the
4342 kernel was under memory pressure. This makes use of the kernel's PSI
4343 interface. Most long-running services in systemd have been hooked up
4344 with this, and in particular systems with low memory should benefit
4347 * Service units gained new settings MemoryPressureWatch= and
4348 MemoryPressureThresholdSec= to configure the PSI memory pressure
4349 logic individually. If these options are used, the
4350 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
4351 variables will be set for the invoked processes to inform them about
4352 the requested memory pressure behaviour. (This is used by the
4353 aforementioned sd-events API additions, if set.)
4355 * systemd-analyze gained a new "malloc" verb that shows the output
4356 generated by glibc's malloc_info() on services that support it. Right
4357 now, only the service manager has been updated accordingly. This
4358 call requires privileges.
4360 User & Session Management:
4362 * The sd-login API gained a new call sd_session_get_username() to
4363 return the user name of the owner of a login session. It also gained
4364 a new call sd_session_get_start_time() to retrieve the time the login
4365 session started. A new call sd_session_get_leader() has been added to
4366 return the PID of the "leader" process of a session. A new call
4367 sd_uid_get_login_time() returns the time since the specified user has
4368 most recently been continuously logged in with at least one session.
4370 * JSON user records gained a new set of fields capabilityAmbientSet and
4371 capabilityBoundingSet which contain a list of POSIX capabilities to
4372 set for the logged in users in the ambient and bounding sets,
4373 respectively. homectl gained the ability to configure these two sets
4374 for users via --capability-bounding-set=/--capability-ambient-set=.
4376 * pam_systemd learnt two new module options
4377 default-capability-bounding-set= and default-capability-ambient-set=,
4378 which configure the default bounding sets for users as they are
4379 logging in, if the JSON user record doesn't specify this explicitly
4380 (see above). The built-in default for the ambient set now contains
4381 the CAP_WAKE_ALARM, thus allowing regular users who may log in
4382 locally to resume from a system suspend via a timer.
4384 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
4385 call to update the TTY of a session after it has been allocated. This
4386 is useful for SSH sessions which are typically allocated first, and
4387 for which a TTY is added later.
4389 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
4390 combines the various other sd_pid_notify() flavours into one: takes a
4391 format string, an overriding PID, and a set of file descriptors to
4392 send. It also gained a new call sd_pid_notify_barrier() call which is
4393 equivalent to sd_notify_barrier() but allows the originating PID to
4396 * "loginctl list-users" and "loginctl list-sessions" will now show the
4397 state of each logged in user/session in their tabular output. It will
4398 also show the current idle state of sessions.
4402 * systemd-dissect will now show the intended CPU architecture of an
4405 * systemd-dissect will now install itself as mount helper for the "ddi"
4406 pseudo-file system type. This means you may now mount DDIs directly
4407 via /bin/mount or /etc/fstab, making full use of embedded Verity
4408 information and all other DDI features.
4410 Example: mount -t ddi myimage.raw /some/where
4412 * The systemd-dissect tool gained the new switches --attach/--detach to
4413 attach/detach a DDI to a loopback block device without mounting it.
4414 It will automatically derive the right sector size from the image
4415 and set up Verity and similar, but not mount the file systems in it.
4417 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
4418 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
4419 implied. Given that these file systems are typically untrusted, this
4420 should make mounting them automatically have less of a security
4423 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
4424 systemd-tmpfiles, …) now understand a new switch --image-policy= which
4425 takes a string encoding image dissection policy. With this mechanism
4426 automatic discovery and use of specific partition types and the
4427 cryptographic requirements on the partitions (Verity, LUKS, …) can be
4428 restricted, permitting better control of the exposed attack surfaces
4429 when mounting disk images. systemd-gpt-auto-generator will honour such
4430 an image policy too, configurable via the systemd.image_policy= kernel
4431 command line option. Unit files gained the RootImagePolicy=,
4432 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
4433 disk images a service runs off.
4435 * systemd-analyze gained a new verb "image-policy" to validate and
4436 parse image policy strings.
4438 * systemd-dissect gained support for a new --validate switch to
4439 superficially validate DDI structure, and check whether a specific
4440 image policy allows the DDI.
4442 * systemd-dissect gained support for a new --mtree-hash switch to
4443 optionally disable calculating mtree hashes, which can be slow on
4446 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
4447 are now able to operate on directories too, other than images.
4451 * networkd's GENEVE support as gained a new .network option
4452 InheritInnerProtocol=.
4454 * The [Tunnel] section in .netdev files has gained a new setting
4455 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
4457 * A new global IPv6PrivacyExtensions= setting has been added that
4458 selects the default value of the per-network setting of the same
4461 * The predictable network interface naming logic was extended to
4462 include SR-IOV-R "representor" information in network interface
4463 names. Unfortunately, this feature was not enabled by default and can
4464 only be enabled at compilation time by setting
4465 -Ddefault-net-naming-scheme=v254.
4467 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
4468 the RFC8910 captive portal option.
4472 * udevadm gained the new "verify" verb for validating udev rules files
4475 * udev gained a new tool "iocost" that can be used to configure QoS IO
4476 cost data based on hwdb information onto suitable block devices. Also
4477 see https://github.com/iocost-benchmark/iocost-benchmarks.
4479 TPM2 Support + Disk Encryption & Authentication:
4481 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
4482 ("Storage Root Key") as first step in the TPM2, and then use that
4483 for binding FDE to, if TPM2 support is used. This matches
4484 recommendations of TCG (see
4485 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
4487 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
4488 understand textual identifiers for these PCRs.
4490 * systemd-veritysetup + /etc/veritytab gained support for a series of
4491 new options: hash-offset=, superblock=, format=, data-block-size=,
4492 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
4493 fec-offset=, fec-roots= to configure various aspects of a Verity
4496 * systemd-cryptsetup + /etc/crypttab gained support for a new
4497 veracrypt-pim= option for setting the Personal Iteration Multiplier
4498 of veracrypt volumes.
4500 * systemd-integritysetup + /etc/integritytab gained support for a new
4501 mode= setting for controlling the dm-integrity mode (journal, bitmap,
4502 direct) for the volume.
4504 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
4505 registers, their symbolic names and current values.
4509 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
4510 access right is specified this is equivalent to "x" but only if the
4511 inode in question already has the executable bit set for at least
4512 some user/group. Otherwise the "x" bit will be turned off.
4514 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
4515 with C+ will result in a "merge" copy, i.e. all files of the source
4516 tree are copied into the target tree, even if that tree already
4517 exists, resulting in a combined tree of files already present in the
4518 target tree and those copied in.
4520 * systemd-tmpfiles gained a new --graceful switch. If specified lines
4521 with unknown users/groups will silently be skipped.
4525 * systemd-notify gained two new options --fd= and --fdname= for sending
4526 arbitrary file descriptors to the service manager (while specifying an
4527 explicit name for it).
4529 * systemd-notify gained a new --exec switch, which makes it execute the
4530 specified command line after sending the requested messages. This is
4531 useful for sending out READY=1 first, and then continuing invocation
4532 without changing process ID, so that the tool can be nicely used
4533 within an ExecStart= line of a unit file that uses Type=notify.
4535 sd-event + sd-bus APIs:
4537 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
4538 which may be used to explicitly end a rate-limit state an event
4539 source might be in, resetting all rate limiting counters.
4541 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
4542 sockets, it will now encode the "description" set via
4543 sd_bus_set_description() into the source socket address. It will also
4544 look for this information when accepting a connection. This is useful
4545 to track individual D-Bus connections on a D-Bus broker for debug
4550 * systemd-resolved gained a new resolved.conf setting
4551 StateRetentionSec= which may be used to retain cached DNS records
4552 even after their nominal TTL, and use them in case upstream DNS
4553 servers cannot be reached. This can be used to make name resolution
4554 more resilient in case of network problems.
4556 * resolvectl gained a new verb "show-cache" to show the current cache
4557 contents of systemd-resolved. This verb communicates with the
4558 systemd-resolved daemon and requires privileges.
4562 * Meson >= 0.60.0 is now required to build systemd.
4564 * The default keymap to apply may now be chosen at build-time via the
4565 new -Ddefault-keymap= meson option.
4567 * Most of systemd's long-running services now have a generic handler of
4568 the SIGRTMIN+18 signal handler which executes various operations
4569 depending on the sigqueue() parameter sent along. For example, values
4570 0x100…0x107 allow changing the maximum log level of such
4571 services. 0x200…0x203 allow changing the log target of such
4572 services. 0x300 make the services trim their memory similarly to the
4573 automatic PSI-triggered action, see above. 0x301 make the services
4574 output their malloc_info() data to the logs.
4576 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
4577 files, inspired by systemctl's verbs of the same name which edit unit
4578 files. Similarly, networkctl gained the same verbs for editing
4579 .network, .netdev, .link files.
4581 * A new syscall filter group "@sandbox" has been added that contains
4582 syscalls for sandboxing system calls such as those for seccomp and
4585 * New documentation has been added:
4587 https://systemd.io/COREDUMP
4588 https://systemd.io/MEMORY_PRESSURE
4591 * systemd-firstboot gained a new --reset option. If specified, the
4592 settings in /etc/ it knows how to initialize are reset.
4594 * systemd-sysext is now a multi-call binary and is also installed under
4595 the systemd-confext alias name (via a symlink). When invoked that way
4596 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
4597 powerful, atomic, secure configuration management of sorts, that
4598 locally can merge configuration from multiple confext configuration
4599 images into a single immutable tree.
4601 * The --network-macvlan=, --network-ipvlan=, --network-interface=
4602 switches of systemd-nspawn may now optionally take the intended
4603 network interface inside the container.
4605 * All our programs will now send an sd_notify() message with their exit
4606 status in the EXIT_STATUS= field when exiting, using the usual
4607 protocol, including PID 1. This is useful for VMMs and container
4608 managers to collect an exit status from a system as it shuts down, as
4609 set via "systemctl exit …". This is particularly useful in test cases
4610 and similar, as invocations via a VM can now nicely propagate an exit
4611 status to the host, similar to local processes.
4613 * systemd-run gained a new switch --expand-environment=no to disable
4614 server-side environment variable expansion in specified command
4615 lines. Expansion defaults to enabled for all execution types except
4616 --scope, where it defaults to off (and prints a warning) for backward
4617 compatibility reasons. --scope will be flipped to enabled by default
4618 too in a future release. If you are using --scope and passing a '$'
4619 character in the payload you should start explicitly using
4620 --expand-environment=yes/no according to the use case.
4622 * The systemd-system-update-generator has been updated to also look for
4623 the special flag file /etc/system-update in addition to the existing
4624 support for /system-update to decide whether to enter system update
4627 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
4628 mount options by default.
4630 * systemd-fstab-generator now understands two new kernel command line
4631 options systemd.mount-extra= and systemd.swap-extra=, which configure
4632 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
4633 will be ran on these block devices, like it already happens for
4634 'root='. It also now supports the new fstab.extra and
4635 fstab.extra.initrd credentials that may contain additional /etc/fstab
4636 lines to apply at boot.
4638 * systemd-getty-generator now understands two new credentials
4639 getty.ttys.container and getty.ttys.serial. These credentials may
4640 contain a list of TTY devices – one per line – to instantiate
4641 container-getty@.service and serial-getty@.service on.
4643 * The getty/serial-getty/container-getty units now import the 'agetty.*'
4644 and 'login.*' credentials, which are consumed by the 'login' and
4645 'agetty' programs starting from util-linux v2.40.
4647 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
4648 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
4649 which case the Path= setting is taken relative to the ESP or XBOOTLDR
4650 partitions, rather than the system's root directory /. The relevant
4651 directories are automatically discovered.
4653 * The systemd-ac-power tool gained a new switch --low, which reports
4654 whether the battery charge is considered "low", similar to how the
4655 s2h suspend logic checks this state to decide whether to enter system
4656 suspend or hibernation.
4658 * The /etc/os-release file can now have two new optional fields
4659 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
4662 * When the system hibernates, information about the device and offset
4663 used is now written to a non-volatile EFI variable. On next boot the
4664 system will attempt to resume from the location indicated in this EFI
4665 variable. This should make hibernation a lot more robust, while
4666 requiring no manual configuration of the resume location.
4668 * The $XDG_STATE_HOME environment variable (added in more recent
4669 versions of the XDG basedir specification) is now honoured to
4670 implement the StateDirectory= setting in user services.
4672 * A new component "systemd-battery-check" has been added. It may run
4673 during early boot (usually in the initrd), and checks the battery
4674 charge level of the system. In case the charge level is very low the
4675 user is notified (graphically via Plymouth – if available – as well
4676 as in text form on the console), and the system is turned off after a
4677 10s delay. The feature can be disabled by passing
4678 systemd.battery_check=0 through the kernel command line.
4680 * The 'passwdqc' library is now supported as an alternative to the
4681 'pwquality' library and can be selected at build time.
4683 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
4684 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
4685 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
4686 Arian van Putten, Arthur Shau, A S Alam,
4687 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
4688 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
4689 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
4690 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
4691 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
4692 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
4693 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
4694 David Schroeder, David Tardon, dependabot[bot],
4695 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
4696 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
4697 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
4698 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
4699 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
4700 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
4701 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
4702 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
4703 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
4704 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
4705 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
4706 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
4707 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
4708 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
4709 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
4710 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
4711 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
4712 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
4713 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
4714 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
4715 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
4716 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
4717 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
4718 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
4719 saikat0511, Samanta Navarro, Sam James, Sam Morris,
4720 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
4721 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
4722 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
4723 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
4724 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
4725 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
4726 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
4727 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
4728 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
4729 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
4730 Дамјан Георгиевски, наб
4732 — Edinburgh, 2023-07-28
4736 Announcements of Future Feature Removals and Incompatible Changes:
4738 * We intend to remove cgroup v1 support from systemd release after the
4739 end of 2023. If you run services that make explicit use of cgroup v1
4740 features (i.e. the "legacy hierarchy" with separate hierarchies for
4741 each controller), please implement compatibility with cgroup v2 (i.e.
4742 the "unified hierarchy") sooner rather than later. Most of Linux
4743 userspace has been ported over already.
4745 * We intend to remove support for split-usr (/usr mounted separately
4746 during boot) and unmerged-usr (parallel directories /bin and
4747 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
4748 half of 2023, in the first release that falls into that time window.
4749 For more details, see:
4750 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
4752 * We intend to change behaviour w.r.t. units of the per-user service
4753 manager and sandboxing options, so that they work without having to
4754 manually enable PrivateUsers= as well, which is not required for
4755 system units. To make this work, we will implicitly enable user
4756 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
4757 user unit. The drawback is that system users will no longer be visible
4758 (and appear as 'nobody') to the user unit when a sandboxing option is
4759 enabled. By definition a sandboxed user unit should run with reduced
4760 privileges, so impact should be small. This will remove a great source
4761 of confusion that has been reported by users over the years, due to
4762 how these options require an extra setting to be manually enabled when
4763 used in the per-user service manager, as opposed as to the system
4764 service manager. We plan to enable this change in the next release
4765 later this year. For more details, see:
4766 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
4768 Deprecations and incompatible changes:
4770 * systemctl will now warn when invoked without /proc/ mounted
4771 (e.g. when invoked after chroot() into an directory tree without the
4772 API mount points like /proc/ being set up.) Operation in such an
4773 environment is not fully supported.
4775 * The return value of 'systemctl is-active|is-enabled|is-failed' for
4776 unknown units is changed: previously 1 or 3 were returned, but now 4
4777 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
4779 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
4780 systemd-hwdb (added in 2014) should be used instead.
4782 * 'bootctl --json' now outputs a single JSON array, instead of a stream
4783 of newline-separated JSON objects.
4785 * Udev rules in 60-evdev.rules have been changed to load hwdb
4786 properties for all modalias patterns. Previously only the first
4787 matching pattern was used. This could change what properties are
4788 assigned if the user has more and less specific patterns that could
4789 match the same device, but it is expected that the change will have
4790 no effect for most users.
4792 * systemd-networkd-wait-online exits successfully when all interfaces
4793 are ready or unmanaged. Previously, if neither '--any' nor
4794 '--interface=' options were used, at least one interface had to be in
4795 configured state. This change allows the case where systemd-networkd
4796 is enabled, but no interfaces are configured, to be handled
4797 gracefully. It may occur in particular when a different network
4798 manager is also enabled and used.
4800 * Some compatibility helpers were dropped: EmergencyAction= in the user
4801 manager, as well as measuring kernel command line into PCR 8 in
4802 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
4805 * The '-Dupdate-helper-user-timeout=' build-time option has been
4806 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
4807 integer as parameter instead of a string.
4809 * The DDI image dissection logic (which backs RootImage= in service
4810 unit files, the --image= switch in various tools such as
4811 systemd-nspawn, as well as systemd-dissect) will now only mount file
4812 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
4813 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
4814 variable. These file systems are fairly well supported and maintained
4815 in current kernels, while others are usually more niche, exotic or
4816 legacy and thus typically do not receive the same level of security
4819 * The default per-link multicast DNS mode is changed to "yes"
4820 (that was previously "no"). As the default global multicast DNS mode
4821 has been "yes" (but can be changed by the build option), now the
4822 multicast DNS is enabled on all links by default. You can disable the
4823 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
4824 or on an interface by calling "resolvectl mdns INTERFACE no".
4828 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
4829 (UKIs) has been added. This replaces functionality provided by
4830 'dracut --uefi' and extends it with automatic calculation of PE file
4831 offsets, insertion of signed PCR policies generated by
4832 systemd-measure, support for initrd concatenation, signing of the
4833 embedded Linux image and the combined image with sbsign, and
4834 heuristics to autodetect the kernel uname and verify the splash
4837 Changes in systemd and units:
4839 * A new service type Type=notify-reload is defined. When such a unit is
4840 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
4841 service process. The manager will then wait until it receives a
4842 "RELOADING=1" followed by a "READY=1" notification from the unit as
4843 response (via sd_notify()). Otherwise, this type is the same as
4844 Type=notify. A new setting ReloadSignal= may be used to change the
4845 signal to send from the default of SIGHUP.
4847 user@.service, systemd-networkd.service, systemd-udevd.service, and
4848 systemd-logind have been updated to this type.
4850 * Initrd environments which are not on a pure memory file system (e.g.
4851 overlayfs combination as opposed to tmpfs) are now supported. With
4852 this change, during the initrd → host transition ("switch root")
4853 systemd will erase all files of the initrd only when the initrd is
4854 backed by a memory file system such as tmpfs.
4856 * New per-unit MemoryZSwapMax= option has been added to configure
4857 memory.zswap.max cgroup properties (the maximum amount of zswap
4860 * A new LogFilterPatterns= option has been added for units. It may be
4861 used to specify accept/deny regular expressions for log messages
4862 generated by the unit, that shall be enforced by systemd-journald.
4863 Rejected messages are neither stored in the journal nor forwarded.
4864 This option may be used to suppress noisy or uninteresting messages
4867 * The manager has a new
4868 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
4869 query process ownership via a PIDFD, which is more resilient against
4870 PID recycling issues.
4872 * Scope units now support OOMPolicy=. Login session scopes default to
4873 OOMPolicy=continue, allowing login scopes to survive the OOM killer
4874 terminating some processes in the scope.
4876 * systemd-fstab-generator now supports x-systemd.makefs option for
4877 /sysroot/ (in the initrd).
4879 * The maximum rate at which daemon reloads are executed can now be
4880 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
4881 options. (Or the equivalent on the kernel command line:
4882 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
4883 addition, systemd now logs the originating unit and PID when a reload
4884 request is received over D-Bus.
4886 * When enabling a swap device systemd will now reinitialize the device
4887 when the page size of the swap space does not match the page size of
4888 the running kernel. Note that this requires the 'swapon' utility to
4889 provide the '--fixpgsz' option, as implemented by util-linux, and it
4890 is not supported by busybox at the time of writing.
4892 * systemd now executes generator programs in a mount namespace
4893 "sandbox" with most of the file system read-only and write access
4894 restricted to the output directories, and with a temporary /tmp/
4895 mount provided. This provides a safeguard against programming errors
4896 in the generators, but also fixes here-docs in shells, which
4897 previously didn't work in early boot when /tmp/ wasn't available
4898 yet. (This feature has no security implications, because the code is
4899 still privileged and can trivially exit the sandbox.)
4901 * The system manager will now parse a new "vmm.notify_socket"
4902 system credential, which may be supplied to a VM via SMBIOS. If
4903 found, the manager will send a "READY=1" notification on the
4904 specified socket after boot is complete. This allows readiness
4905 notification to be sent from a VM guest to the VM host over a VSOCK
4908 * The sample PAM configuration file for systemd-user@.service now
4909 includes a call to pam_namespace. This puts children of user@.service
4910 in the expected namespace. (Many distributions replace their file
4911 with something custom, so this change has limited effect.)
4913 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
4914 can be used to override the mount units burst late limit for
4915 parsing '/proc/self/mountinfo', which was introduced in v249.
4918 * Drop-ins for init.scope changing control group resource limits are
4919 now applied, while they were previously ignored.
4921 * New build-time configuration options '-Ddefault-timeout-sec=' and
4922 '-Ddefault-user-timeout-sec=' have been added, to let distributions
4923 choose the default timeout for starting/stopping/aborting system and
4924 user units respectively.
4926 * Service units gained a new setting OpenFile= which may be used to
4927 open arbitrary files in the file system (or connect to arbitrary
4928 AF_UNIX sockets in the file system), and pass the open file
4929 descriptor to the invoked process via the usual file descriptor
4930 passing protocol. This is useful to give unprivileged services access
4931 to select files which have restrictive access modes that would
4932 normally not allow this. It's also useful in case RootDirectory= or
4933 RootImage= is used to allow access to files from the host environment
4934 (which is after all not visible from the service if these two options
4939 * The new net naming scheme "v253" has been introduced. In the new
4940 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
4941 a PCI bus. This extends the coverage of predictable interface names
4942 in some embedded systems.
4944 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
4945 a more informative path on some embedded systems.
4947 * Partition block devices will now also get symlinks in
4948 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
4949 block device nodes via the kernel's "diskseq" value. Previously those
4950 symlinks were only created for the main block device.
4952 * A new operator '-=' is supported for SYMLINK variables. This allows
4953 symlinks to be unconfigured even if an earlier rule added them.
4955 * 'udevadm --trigger --settle' now also works for network devices
4956 that are being renamed.
4958 Changes in sd-boot, bootctl, and the Boot Loader Specification:
4960 * systemd-boot now passes its random seed directly to the kernel's RNG
4961 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
4962 means the RNG gets seeded very early in boot before userspace has
4965 * systemd-boot will pass a disk-backed random seed – even when secure
4966 boot is enabled – if it can additionally get a random seed from EFI
4967 itself (via EFI's RNG protocol), or a prior seed in
4968 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
4970 * systemd-boot-system-token.service was renamed to
4971 systemd-boot-random-seed.service and extended to always save a random
4972 seed to ESP on every boot when a compatible boot loader is used. This
4973 allows a refreshed random seed to be used in the boot loader.
4975 * systemd-boot handles various seed inputs using a domain- and
4976 field-separated hashing scheme.
4978 * systemd-boot's 'random-seed-mode' option has been removed. A system
4979 token is now always required to be present for random seeds to be
4982 * systemd-boot now supports being loaded from other locations than the
4983 ESP, for example for direct kernel boot under QEMU or when embedded
4986 * systemd-boot now parses SMBIOS information to detect
4987 virtualization. This information is used to skip some warnings which
4988 are not useful in a VM and to conditionalize other aspects of
4991 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
4992 Secure Boot automated certificate enrollment from the ESP only if it
4993 is considered 'safe' to do so. At the moment 'safe' means running in
4996 * systemd-stub now processes random seeds in the same way as
4997 systemd-boot already does, in case a unified kernel image is being
4998 used from a different bootloader than systemd-boot, or without any
5001 * bootctl will now generate a system token on all EFI systems, even
5002 virtualized ones, and is activated in the case that the system token
5003 is missing from either sd-boot and sd-stub booted systems.
5005 * bootctl now implements two new verbs: 'kernel-identify' prints the
5006 type of a kernel image file, and 'kernel-inspect' provides
5007 information about the embedded command line and kernel version of
5010 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
5011 as for kernel-install.
5013 * The JSON output of "bootctl list" will now contain two more fields:
5014 isDefault and isSelected are boolean fields set to true on the
5015 default and currently booted boot menu entries.
5017 * bootctl gained a new verb "unlink" for removing a boot loader entry
5018 type #1 file from disk in a safe and robust way.
5020 * bootctl also gained a new verb "cleanup" that automatically removes
5021 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
5022 is not referenced anymore by any installed Type #1 boot loader
5023 specification entry. This is particularly useful in environments where
5024 a large number of entries reference the same or partly the same
5025 resources (for example, for snapshot-based setups).
5027 Changes in kernel-install:
5029 * A new "installation layout" can be configured as layout=uki. With
5030 this setting, a Boot Loader Specification Type#1 entry will not be
5031 created. Instead, a new kernel-install plugin 90-uki-copy.install
5032 will copy any .efi files from the staging area into the boot
5033 partition. A plugin to generate the UKI .efi file must be provided
5036 Changes in systemctl:
5038 * 'systemctl reboot' has dropped support for accepting a positional
5039 argument as the argument to the reboot(2) syscall. Please use the
5040 --reboot-argument= option instead.
5042 * 'systemctl disable' will now warn when called on units without
5043 install information. A new --no-warn option has been added that
5044 silences this warning.
5046 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
5047 of the drop-in to edit. (Previously, 'override.conf' was always
5050 * 'systemctl list-dependencies' now respects --type= and --state=.
5052 * 'systemctl kexec' now supports XEN VMM environments.
5054 * 'systemctl edit' will now tell the invoked editor to jump into the
5055 first line with actual unit file data, skipping over synthesized
5058 Changes in systemd-networkd and related tools:
5060 * The [DHCPv4] section in .network file gained new SocketPriority=
5061 setting that assigns the Linux socket priority used by the DHCPv4 raw
5062 socket. This may be used in conjunction with the
5063 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
5064 desired ethernet 802.1Q frame priority for DHCPv4 initial
5065 packets. This cannot be achieved with netfilter mangle tables because
5066 of the raw socket bypass.
5068 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
5069 new QuickAck= boolean setting that enables the TCP quick ACK mode for
5070 the routes configured by the acquired DHCPv4 lease or received router
5071 advertisements (RAs).
5073 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
5074 routes) now accepts three values, for high, medium, and low preference
5075 of the router (which can be set with the RouterPreference=) setting.
5077 * systemd-networkd-wait-online now supports matching via alternative
5080 * The [DHCPv6] section in .network file gained new SendRelease=
5081 setting which enables the DHCPv6 client to send release when
5082 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
5083 It is enabled by default.
5085 * If the Address= setting in [Network] or [Address] sections in .network
5086 specified without its prefix length, then now systemd-networkd assumes
5087 /32 for IPv4 or /128 for IPv6 addresses.
5089 * networkctl shows network and link file dropins in status output.
5091 Changes in systemd-dissect:
5093 * systemd-dissect gained a new option --list, to print the paths of
5094 all files and directories in a DDI.
5096 * systemd-dissect gained a new option --mtree, to generate a file
5097 manifest compatible with BSD mtree(5) of a DDI
5099 * systemd-dissect gained a new option --with, to execute a command with
5100 the specified DDI temporarily mounted and used as working
5101 directory. This is for example useful to convert a DDI to "tar"
5102 simply by running it within a "systemd-dissect --with" invocation.
5104 * systemd-dissect gained a new option --discover, to search for
5105 Discoverable Disk Images (DDIs) in well-known directories of the
5106 system. This will list machine, portable service and system extension
5109 * systemd-dissect now understands 2nd stage initrd images stored as a
5110 Discoverable Disk Image (DDI).
5112 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
5113 disk UUID stored in the GPT header) among the other data it can show.
5115 * systemd-dissect gained a new --in-memory switch to operate on an
5116 in-memory copy of the specified DDI file. This is useful to access a
5117 DDI with write access without persisting any changes. It's also
5118 useful for accessing a DDI without keeping the originating file
5121 * The DDI dissection logic will now automatically detect the intended
5122 sector size of disk images stored in files, based on the GPT
5123 partition table arrangement. Loopback block devices for such DDIs
5124 will then be configured automatically for the right sector size. This
5125 is useful to make dealing with modern 4K sector size DDIs fully
5126 automatic. The systemd-dissect tool will now show the detected sector
5127 size among the other DDI information in its output.
5129 Changes in systemd-repart:
5131 * systemd-repart gained new options --include-partitions= and
5132 --exclude-partitions= to filter operation on partitions by type UUID.
5133 This allows systemd-repart to be used to build images in which the
5134 type of one partition is set based on the contents of another
5135 partition (for example when the boot partition shall include a verity
5136 hash of the root partition).
5138 * systemd-repart also gained a --defer-partitions= option that is
5139 similar to --exclude-partitions=, but the size of the partition is
5140 still taken into account when sizing partitions, but without
5143 * systemd-repart gained a new --sector-size= option to specify what
5144 sector size should be used when an image is created.
5146 * systemd-repart now supports generating erofs file systems via
5147 CopyFiles= (a read-only file system similar to squashfs).
5149 * The Minimize= option was extended to accept "best" (which means the
5150 most minimal image possible, but may require multiple attempts) and
5151 "guess" (which means a reasonably small image).
5153 * The systemd-growfs binary now comes with a regular unit file template
5154 systemd-growfs@.service which can be instantiated directly for any
5155 desired file system. (Previously, the unit was generated dynamically
5156 by various generators, but no regular unit file template was
5159 Changes in journal tools:
5161 * Various systemd tools will append extra fields to log messages when
5162 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
5163 this includes information about D-Bus messages when sd-bus is used,
5164 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
5165 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
5166 Details of what is logged and when are subject to change.
5168 * The systemd-journald-audit.socket can now be disabled via the usual
5169 "systemctl disable" mechanism to stop collection of audit
5170 messages. Please note that it is not enabled statically anymore and
5171 must be handled by the preset/enablement logic in package
5172 installation scripts.
5174 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
5175 be used to curtail disk use by systemd-journal-remote. This is
5176 similar to the options supported by systemd-journald.
5178 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
5181 * When enrolling new keys systemd-cryptenroll now supports unlocking
5182 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
5183 password was strictly required to be specified.
5185 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
5186 (except for tokens with user verification, UV) to identify tokens
5187 before authentication. Multiple FIDO2 tokens can now be enrolled at
5188 the same time, and systemd-cryptsetup will automatically select one
5189 that corresponds to one of the available LUKS key slots.
5191 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
5192 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
5193 bank and number into which the volume key should be measured. This is
5194 automatically enabled for the encrypted root volume discovered and
5195 activated by systemd-gpt-auto-generator.
5197 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
5198 "noexec,nosuid,nodev".
5200 * systemd-gpt-auto-generator will now honour the rootfstype= and
5201 rootflags= kernel command line switches for root file systems it
5202 discovers, to match behaviour in case an explicit root fs is
5203 specified via root=.
5205 * systemd-pcrphase gained new options --machine-id and --file-system=
5206 to measure the machine-id and mount point information into PCR 15.
5207 New service unit files systemd-pcrmachine.service and
5208 systemd-pcrfs@.service have been added that invoke the tool with
5209 these switches during early boot.
5211 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
5212 with a success exit code even if no TPM device is detected.
5214 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
5215 making it harder to brute-force.
5217 Changes in other tools:
5219 * systemd-homed gained support for luksPbkdfForceIterations (the
5220 intended number of iterations for the PBKDF operation on LUKS).
5222 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
5223 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
5224 may now be used to specify additional arguments for mkfs when
5225 systemd-homed formats a file system.
5227 * systemd-hostnamed now exports the contents of
5228 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
5229 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
5230 unprivileged code to access those values.
5232 systemd-hostnamed also exports the SUPPORT_END= field from
5233 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
5234 this to show the status of the installed system.
5236 * systemd-measure gained an --append= option to sign multiple phase
5237 paths with different signing keys. This allows secrets to be
5238 accessible only in certain parts of the boot sequence. Note that
5239 'ukify' provides similar functionality in a more accessible form.
5241 * systemd-timesyncd will now write a structured log message with
5242 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
5243 on a on-disk timestamp, similarly to what it did when reaching
5244 synchronization via NTP.
5246 * systemd-timesyncd will now update the on-disk timestamp file on each
5247 boot at least once, making it more likely that the system time
5248 increases in subsequent boots.
5250 * systemd-vconsole-setup gained support for system/service credentials:
5251 vconsole.keymap/vconsole.keymap_toggle and
5252 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
5253 the similarly-named options in vconsole.conf.
5255 * systemd-localed will now save the XKB keyboard configuration to
5256 /etc/vconsole.conf, and also read it from there with a higher
5257 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
5258 file. Previously, this information was stored in the former file in
5259 converted form, and only in latter file in the original form. Tools
5260 which want to access keyboard configuration can now do so from a
5263 * systemd-resolved gained support for configuring the nameservers and
5264 search domains via kernel command line (nameserver=, domain=) and
5265 credentials (network.dns, network.search_domains).
5267 * systemd-resolved will now synthesize host names for the DNS stub
5268 addresses it supports. Specifically when "_localdnsstub" is resolved,
5269 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
5270 127.0.0.54 is returned.
5272 * systemd-notify will now send a "RELOADING=1" notification when called
5273 with --reloading, and "STOPPING=1" when called with --stopping. This
5274 can be used to implement notifications from units where it's easier
5275 to call a program than to use the sd-daemon library.
5277 * systemd-analyze's 'plot' command can now output its information in
5278 JSON, controlled via the --json= switch. Also, new --table, and
5279 --no-legend options have been added.
5281 * 'machinectl enable' will now automatically enable machines.target
5282 unit in addition to adding the machine unit to the target.
5284 Similarly, 'machinectl start|stop' gained a --now option to enable or
5285 disable the machine unit when starting or stopping it.
5287 * systemd-sysusers will now create /etc/ if it is missing.
5289 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
5290 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
5291 added to provide the new initial value for the new automated battery
5292 estimation functionality. If 'HibernateDelaySec=' is set to any value,
5293 the automated estimate (and thus the automated hibernation on low
5294 battery to avoid data loss) functionality will be disabled.
5296 * Default tmpfiles.d/ configuration will now automatically create
5297 credentials storage directory '/etc/credstore/' with the appropriate,
5298 secure permissions. If '/run/credstore/' exists, its permissions will
5299 be fixed too in case they are not correct.
5301 Changes in libsystemd and shared code:
5303 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
5304 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
5306 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
5307 128-bit ID in files such as /etc/machine-id has an invalid
5308 format. They also accept NULL as output parameter in more places,
5309 which is useful when the caller only wants to validate the inputs and
5310 does not need the output value.
5312 * sd-login gained new functions sd_pidfd_get_session(),
5313 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
5314 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
5315 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
5316 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
5317 but accept a PIDFD instead of a PID.
5319 * sd-path (and systemd-path) now export four new paths:
5320 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
5321 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
5322 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
5323 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
5325 * sd_notify() now supports AF_VSOCK as transport for notification
5326 messages (in addition to the existing AF_UNIX support). This is
5327 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
5329 * Detection of chroot() environments now works if /proc/ is not
5330 mounted. This affects systemd-detect-virt --chroot, but also means
5331 that systemd tools will silently skip various operations in such an
5334 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
5335 virtualization is now detected.
5337 Changes in the build system:
5339 * Standalone variants of systemd-repart and systemd-shutdown may now be
5340 built (if -Dstandalone=true).
5342 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
5343 example, allow scripts to conditionalize execution on AC power
5346 * The libp11kit library is now loaded through dlopen(3).
5348 Changes in the documentation:
5350 * Specifications that are not closely tied to systemd have moved to
5351 https://uapi-group.org/specifications/: the Boot Loader Specification
5352 and the Discoverable Partitions Specification.
5354 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
5355 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
5356 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
5357 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
5358 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
5359 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
5360 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
5361 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
5362 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
5363 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
5364 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
5365 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
5366 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
5367 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
5368 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
5369 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
5370 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
5371 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
5372 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
5373 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
5374 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
5375 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
5376 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
5377 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
5378 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
5379 reuben olinsky, Richard E. van der Luit, Richard Phibel,
5380 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
5381 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
5382 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
5383 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
5384 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
5385 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
5386 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
5387 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
5390 — Warsaw, 2023-02-15
5394 Announcements of Future Feature Removals:
5396 * We intend to remove cgroup v1 support from systemd release after the
5397 end of 2023. If you run services that make explicit use of cgroup v1
5398 features (i.e. the "legacy hierarchy" with separate hierarchies for
5399 each controller), please implement compatibility with cgroup v2 (i.e.
5400 the "unified hierarchy") sooner rather than later. Most of Linux
5401 userspace has been ported over already.
5403 * We intend to remove support for split-usr (/usr mounted separately
5404 during boot) and unmerged-usr (parallel directories /bin and
5405 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
5406 half of 2023, in the first release that falls into that time window.
5407 For more details, see:
5408 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
5410 Compatibility Breaks:
5412 * ConditionKernelVersion= checks that use the '=' or '!=' operators
5413 will now do simple string comparisons (instead of version comparisons
5414 à la stverscmp()). Version comparisons are still done for the
5415 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
5416 specified, a shell-style glob match is now done. This creates a minor
5417 incompatibility compared to older systemd versions when the '*', '?',
5418 '[', ']' characters are used, as these will now match as shell globs
5419 instead of literally. Given that kernel version strings typically do
5420 not include these characters we expect little breakage through this
5423 * The service manager will now read the SELinux label used for SELinux
5424 access checks from the unit file at the time it loads the file.
5425 Previously, the label would be read at the moment of the access
5426 check, which was problematic since at that time the unit file might
5427 already have been updated or removed.
5431 * systemd-measure is a new tool for calculating and signing expected
5432 TPM2 PCR values for a given unified kernel image (UKI) booted via
5433 sd-stub. The public key used for the signature and the signed
5434 expected PCR information can be embedded inside the UKI. This
5435 information can be extracted from the UKI by external tools and code
5436 in the image itself and is made available to userspace in the booted
5439 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
5440 updated to make use of this information if available in the booted
5441 kernel: when locking an encrypted volume/credential to the TPM
5442 systemd-cryptenroll/systemd-creds will use the public key to bind the
5443 volume/credential to any kernel that carries PCR information signed
5444 by the same key pair. When unlocking such volumes/credentials
5445 systemd-cryptsetup/systemd-creds will use the signature embedded in
5446 the booted UKI to gain access.
5448 Binding TPM-based disk encryption to public keys/signatures of PCR
5449 values — instead of literal PCR values — addresses the inherent
5450 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
5451 disks remain accessible even if the UKI is updated, without any TPM
5452 specific preparation during the OS update — as long as each UKI
5453 carries the necessary PCR signature information.
5455 Net effect: if you boot a properly prepared kernel, TPM-bound disk
5456 encryption now defaults to be locked to kernels which carry PCR
5457 signatures from the same key pair. Example: if a hypothetical distro
5458 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
5459 by default – bound to only FooOS kernels, and encrypted volumes bound
5460 to the TPM cannot be unlocked on kernels from other sources. (But do
5461 note this behaviour requires preparation/enabling in the UKI, and of
5462 course users can always enroll non-TPM ways to unlock the volume.)
5464 * systemd-pcrphase is a new tool that is invoked at six places during
5465 system runtime, and measures additional words into TPM2 PCR 11, to
5466 mark milestones of the boot process. This allows binding access to
5467 specific TPM2-encrypted secrets to specific phases of the boot
5468 process. (Example: LUKS2 disk encryption key only accessible in the
5469 initrd, but not later.)
5471 Changes in systemd itself, i.e. the manager and units
5473 * The cpu controller is delegated to user manager units by default, and
5474 CPUWeight= settings are applied to the top-level user slice units
5475 (app.slice, background.slice, session.slice). This provides a degree
5476 of resource isolation between different user services competing for
5479 * Systemd can optionally do a full preset in the "first boot" condition
5480 (instead of just enable-only). This behaviour is controlled by the
5481 compile-time option -Dfirst-boot-full-preset. Right now it defaults
5482 to 'false', but the plan is to switch it to 'true' for the subsequent
5485 * Drop-ins are now allowed for transient units too.
5487 * Systemd will set the taint flag 'support-ended' if it detects that
5488 the OS image is past its end-of-support date. This date is declared
5489 in a new /etc/os-release field SUPPORT_END= described below.
5491 * Two new settings ConditionCredential= and AssertCredential= can be
5492 used to skip or fail units if a certain system credential is not
5495 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
5497 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
5498 specify the SMACK security label to use when not specified in a unit
5501 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
5502 specify the default timeout when waiting for device units to
5505 * C.UTF-8 is used as the default locale if nothing else has been
5508 * [Condition|Assert]Firmware= have been extended to support certain
5509 SMBIOS fields. For example
5511 ConditionFirmware=smbios-field(board_name = "Custom Board")
5513 conditionalizes the unit to run only when
5514 /sys/class/dmi/id/board_name contains "Custom Board" (without the
5517 * ConditionFirstBoot= now correctly evaluates as true only during the
5518 boot phase of the first boot. A unit executed later, after booting
5519 has completed, will no longer evaluate this condition as true.
5521 * Socket units will now create sockets in the SELinuxContext= of the
5522 associated service unit, if any.
5524 * Boot phase transitions (start initrd → exit initrd → boot complete →
5525 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
5526 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
5527 unsealed only in the initrd.
5529 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
5530 also be provided to ExecStartPre= processes.
5532 * Various units are now correctly ordered against
5533 initrd-switch-root.target where previously a conflict without
5534 ordering was configured. A stop job for those units would be queued,
5535 but without the ordering it could be executed only after
5536 initrd-switch-root.service, leading to units not being restarted in
5537 the host system as expected.
5539 * In order to fully support the IPMI watchdog driver, which has not yet
5540 been ported to the new common watchdog device interface,
5541 /dev/watchdog0 will be tried first and systemd will silently fallback
5542 to /dev/watchdog if it is not found.
5544 * New watchdog-related D-Bus properties are now published by systemd:
5545 WatchdogDevice, WatchdogLastPingTimestamp,
5546 WatchdogLastPingTimestampMonotonic.
5548 * At shutdown, API virtual files systems (proc, sys, etc.) will be
5551 * At shutdown, systemd will now log about processes blocking unmounting
5554 * A new meson build option 'clock-valid-range-usec-max' was added to
5555 allow disabling system time correction if RTC returns a timestamp far
5558 * Propagated restart jobs will no longer be discarded while a unit is
5561 * PID 1 will now import system credentials from SMBIOS Type 11 fields
5562 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
5563 simple, fast and generic path for supplying credentials to a VM,
5564 without involving external tools such as cloud-init/ignition.
5566 * The CPUWeight= setting of unit files now accepts a new special value
5567 "idle", which configures "idle" level scheduling for the unit.
5569 * Service processes that are activated due to a .timer or .path unit
5570 triggering will now receive information about this via environment
5571 variables. Note that this is information is lossy, as activation
5572 might be coalesced and only one of the activating triggers will be
5573 reported. This is hence more suited for debugging or tracing rather
5574 than for behaviour decisions.
5576 * The riscv_flush_icache(2) system call has been added to the list of
5577 system calls allowed by default when SystemCallFilter= is used.
5579 * The selinux context derived from the target executable, instead of
5580 'init_t' used for the manager itself, is now used when creating
5581 listening sockets for units that specify SELinuxContextFromNet=yes.
5583 Changes in sd-boot, bootctl, and the Boot Loader Specification:
5585 * The Boot Loader Specification has been cleaned up and clarified.
5586 Various corner cases in version string comparisons have been fixed
5587 (e.g. comparisons for empty strings). Boot counting is now part of
5588 the main specification.
5590 * New PCRs measurements are performed during boot: PCR 11 for the
5591 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
5592 took place this is now reported to userspace via the new
5593 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
5595 * As before, systemd-stub will measure kernel parameters and system
5596 credentials into PCR 12. It will now report this fact via the
5597 StubPcrKernelParameters EFI variable to userspace.
5599 * The UEFI monotonic boot counter is now included in the updated random
5600 seed file maintained by sd-boot, providing some additional entropy.
5602 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
5603 of arranging the image manually and jumping to the kernel entry
5604 point. sd-stub also installs a temporary UEFI SecurityOverride to
5605 allow the (unsigned) nested image to be booted. This is safe because
5606 the outer (signed) stub+kernel binary must have been verified before
5607 the stub was executed.
5609 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
5610 is now supported by sd-boot.
5612 * bootctl gained a bunch of new options: --all-architectures to install
5613 binaries for all supported EFI architectures, --root= and --image=
5614 options to operate on a directory or disk image, and
5615 --install-source= to specify the source for binaries to install,
5616 --efi-boot-option-description= to control the name of the boot entry.
5618 * The sd-boot stub exports a StubFeatures flag, which is used by
5619 bootctl to show features supported by the stub that was used to boot.
5621 * The PE section offsets that are used by tools that assemble unified
5622 kernel images have historically been hard-coded. This may lead to
5623 overlapping PE sections which may break on boot. The UKI will now try
5624 to detect and warn about this.
5626 Any tools that assemble UKIs must update to calculate these offsets
5627 dynamically. Future sd-stub versions may use offsets that will not
5628 work with the currently used set of hard-coded offsets!
5630 * sd-stub now accepts (and passes to the initrd and then to the full
5631 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
5632 signatures of expected PCR values, to allow sealing secrets via the
5633 TPM2 against pre-calculated PCR measurements.
5635 Changes in the hardware database:
5637 * 'systemd-hwdb query' now supports the --root= option.
5639 Changes in systemctl:
5641 * systemctl now supports --state= and --type= options for the 'show'
5644 * systemctl gained a new verb 'list-automounts' to list automount
5647 * systemctl gained support for a new --image= switch to be able to
5648 operate on the specified disk image (similar to the existing --root=
5649 which operates relative to some directory).
5651 Changes in systemd-networkd:
5653 * networkd can set Linux NetLabel labels for integration with the
5654 network control in security modules via a new NetLabel= option.
5656 * The RapidCommit= is (re-)introduced to enable faster configuration
5657 via DHCPv6 (RFC 3315).
5659 * networkd gained a new option TCPCongestionControlAlgorithm= that
5660 allows setting a per-route TCP algorithm.
5662 * networkd gained a new option KeepFileDescriptor= to allow keeping a
5663 reference (file descriptor) open on TUN/TAP interfaces, which is
5664 useful to avoid link flaps while the underlying service providing the
5665 interface is being serviced.
5667 * RouteTable= now also accepts route table names.
5669 Changes in systemd-nspawn:
5671 * The --bind= and --overlay= options now support relative paths.
5673 * The --bind= option now supports a 'rootidmap' value, which will
5674 use id-mapped mounts to map the root user inside the container to the
5675 owner of the mounted directory on the host.
5677 Changes in systemd-resolved:
5679 * systemd-resolved now persists DNSOverTLS in its state file too. This
5680 fixes a problem when used in combination with NetworkManager, which
5681 sends the setting only once, causing it to be lost if resolved was
5682 restarted at any point.
5684 * systemd-resolved now exposes a Varlink socket at
5685 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
5686 root. Processed DNS requests in a JSON format will be published to
5687 any clients connected to this socket.
5689 resolvectl gained a 'monitor' verb to make use of this.
5691 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
5692 instead of returning SERVFAIL, as per RFC:
5693 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
5695 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
5696 is still supported.)
5698 Changes in libsystemd and other libraries:
5700 * libsystemd now exports sd_bus_error_setfv() (a convenience function
5701 for setting bus errors), sd_id128_string_equal (a convenience
5702 function for 128-bit ID string comparisons), and
5703 sd_bus_message_read_strv_extend() (a function to incrementally read
5706 * libsystemd now exports sd_device_get_child_first()/_next() as a
5707 high-level interface for enumerating child devices. It also supports
5708 sd_device_new_child() for opening a child device given a device
5711 * libsystemd now exports sd_device_monitor_set()/get_description()
5712 which allow setting a custom description that will be used in log
5713 messages by sd_device_monitor*.
5715 * Private shared libraries (libsystemd-shared-nnn.so,
5716 libsystemd-core-nnn.so) are now installed into arch-specific
5717 directories to allow multi-arch installs.
5719 * A new sd-gpt.h header is now published, listing GUIDs from the
5720 Discoverable Partitions specification. For more details see:
5721 https://systemd.io/DISCOVERABLE_PARTITIONS/
5723 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
5724 database given an explicit path to the file.
5726 * The signal number argument to sd_event_add_signal() now can now be
5727 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
5728 be automatically invoked to block the specified signal. This is
5729 useful to simplify invocations as the caller doesn't have to do this
5732 * A new convenience call sd_event_set_signal_exit() has been added to
5733 sd-event to set up signal handling so that the event loop
5734 automatically terminates cleanly on SIGTERM/SIGINT.
5736 Changes in other components:
5738 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
5739 can now be provided via the credential mechanism.
5741 * systemd-analyze gained a new verb 'compare-versions' that implements
5742 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
5743 'dpkg --compare-versions').
5745 * 'systemd-analyze dump' is extended to accept glob patterns for unit
5746 names to limit the output to matching units.
5748 * tmpfiles.d/ lines can read file contents to write from a credential.
5749 The new modifier char '^' is used to specify that the argument is a
5750 credential name. This mechanism is used to automatically populate
5751 /etc/motd, /etc/issue, and /etc/hosts from credentials.
5753 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
5754 an inode if the specification is prefixed with ':' and the inode
5757 * Default tmpfiles.d/ configuration now carries a line to automatically
5758 use an 'ssh.authorized_keys.root' credential if provided to set up
5759 the SSH authorized_keys file for the root user.
5761 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
5764 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
5765 in base64. This is useful to write arbitrary binary data into files.
5767 * The pkgconfig and rpm macros files now export the directory for user
5768 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
5770 * Detection of Apple Virtualization and detection of Parallels and
5771 KubeVirt virtualization on non-x86 archs have been added.
5773 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
5774 user when their system will become unsupported.
5776 * When performing suspend-then-hibernate, the system will estimate the
5777 discharge rate and use that to set the delay until hibernation and
5778 hibernate immediately instead of suspending when running from a
5779 battery and the capacity is below 5%.
5781 * systemd-sysctl gained a --strict option to fail when a sysctl
5782 setting is unknown to the kernel.
5784 * machinectl supports --force for the 'copy-to' and 'copy-from'
5787 * coredumpctl gained the --root and --image options to look for journal
5788 files under the specified root directory, image, or block device.
5790 * 'journalctl -o' and similar commands now implement a new output mode
5791 "short-delta". It is similar to "short-monotonic", but also shows the
5792 time delta between subsequent messages.
5794 * journalctl now respects the --quiet flag when verifying consistency
5797 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
5798 will indicate whether a message was logged in the 'initrd' phase or
5799 in the 'system' phase of the boot process.
5801 * Journal files gained a new compatibility flag
5802 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
5803 to the storage format that allow reducing size on disk. As with other
5804 compatibility flags, older journalctl versions will not be able to
5805 read journal files using this new format. The environment variable
5806 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
5807 disable this functionality. It is enabled by default.
5809 * systemd-run's --working-directory= switch now works when used in
5810 combination with --scope.
5812 * portablectl gained a --force flag to skip certain sanity checks. This
5813 is implemented using new flags accepted by systemd-portabled for the
5814 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
5815 flag now means that the attach/detach checks whether the units are
5816 already present and running will be skipped. Similarly,
5817 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
5818 image name matches the name declared inside of the image will be
5819 skipped. Callers must be sure to do those checks themselves if
5822 * systemd-portabled will now use the original filename to check
5823 extension-release.NAME for correctness, in case it is passed a
5826 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
5829 * sysext's extension-release files now support '_any' as a special
5830 value for the ID= field, to allow distribution-independent extensions
5831 (e.g.: fully statically compiled binaries, scripts). It also gained
5832 support for a new ARCHITECTURE= field that may be used to explicitly
5833 restrict an image to hosts of a specific architecture.
5835 * systemd-repart now supports creating squashfs partitions. This
5836 requires mksquashfs from squashfs-tools.
5838 * systemd-repart gained a --split flag to also generate split
5839 artifacts, i.e. a separate file for each partition. This is useful in
5840 conjunction with systemd-sysupdate or other tools, or to generate
5841 split dm-verity artifacts.
5843 * systemd-repart is now able to generate dm-verity partitions, including
5846 * systemd-repart can now set a partition UUID to zero, allowing it to
5847 be filled in later, such as when using verity partitions.
5849 * systemd-repart now supports drop-ins for its configuration files.
5851 * Package metadata logged by systemd-coredump in the system journal is
5854 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
5856 * systemd-oomd now automatically links against libatomic, if available.
5858 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
5861 * scope units now also provide oom-kill status.
5863 * systemd-pstore will now try to load only the efi_pstore kernel module
5864 before running, ensuring that pstore can be used.
5866 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
5867 session after a preconfigure timeout.
5869 * systemd-homed will now wait up to 30 seconds for workers to terminate,
5870 rather than indefinitely.
5872 * homectl gained a new '--luks-sector-size=' flag that allows users to
5873 select the preferred LUKS sector size. Must be a power of 2 between 512
5874 and 4096. systemd-userdbd records gained a corresponding field.
5876 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
5877 variable when generating the 'sp_lstchg' field, to ensure an image
5878 build can be reproducible.
5880 * 'udevadm wait' will now listen to kernel uevents too when called with
5883 * When naming network devices udev will now consult the DeviceTree
5884 "alias" fields for the device.
5886 * systemd-udev will now create infiniband/by-path and
5887 infiniband/by-ibdev links for Infiniband verbs devices.
5889 * systemd-udev-trigger.service will now also prioritize input devices.
5891 * ConditionACPower= and systemd-ac-power will now assume the system is
5892 running on AC power if no battery can be found.
5894 * All features and tools using the TPM2 will now communicate with it
5895 using a bind key. Beforehand, the tpm2 support used encrypted sessions
5896 by creating a primary key that was used to encrypt traffic. This
5897 creates a problem as the key created for encrypting the traffic could
5898 be faked by an active interposer on the bus. In cases when a pin is
5899 used, a bind key will be used. The pin is used as the auth value for
5900 the seal key, aka the disk encryption key, and that auth value will be
5901 used in the session establishment. An attacker would need the pin
5902 value to create the secure session and thus an active interposer
5903 without the pin cannot interpose on TPM2 traffic.
5905 * systemd-growfs no longer requires udev to run.
5907 * systemd-backlight now will better support systems with multiple
5910 * systemd-cryptsetup's keyfile-timeout= option now also works when a
5911 device is used as a keyfile.
5913 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
5914 unlocking key from a key file (instead of prompting the user). Note
5915 that this is the key for unlocking the volume in order to be able to
5916 enroll a new key, but it is not the key that is enrolled.
5918 * systemd-dissect gained a new --umount switch that will safely and
5919 synchronously unmount all partitions of an image previously mounted
5920 with 'systemd-dissect --mount'.
5922 * When using gcrypt, all systemd tools and services will now configure
5923 it to prefer the OS random number generator if present.
5925 * All example code shipped with documentation has been relicensed from CC0
5928 * Unit tests will no longer fail when running on a system without
5931 Experimental features:
5933 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
5934 and bpftool >= 7.0).
5936 * sd-boot can automatically enroll SecureBoot keys from files found on
5937 the ESP. This enrollment can be either automatic ('force' mode) or
5938 controlled by the user ('manual' mode). It is sufficient to place the
5939 SecureBoot keys in the right place in the ESP and they will be picked
5940 up by sd-boot and shown in the boot menu.
5942 * The mkosi config in systemd gained support for automatically
5943 compiling a kernel with the configuration appropriate for testing
5944 systemd. This may be useful when developing or testing systemd in
5945 tandem with the kernel.
5947 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
5948 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
5949 Alexander Graf, Alexander Shopov, Alexander Wilson,
5950 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
5951 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
5952 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
5953 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
5954 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
5955 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
5956 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
5957 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
5958 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
5959 David Jaša, David Rheinsberg, David Seifert, David Tardon,
5960 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
5961 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
5962 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
5963 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
5964 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
5965 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
5966 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
5967 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
5968 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
5969 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
5970 JeroenHD, jiangchuangang, João Loureiro,
5971 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
5972 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
5973 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
5974 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
5975 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
5976 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
5977 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
5978 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
5979 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
5980 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
5981 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
5982 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
5983 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
5984 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
5985 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
5986 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
5987 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
5988 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
5989 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
5990 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
5991 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
5992 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
5993 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
5994 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
5996 – The Great Beyond, 2022-10-31 👻
6000 Backwards-incompatible changes:
6002 * The minimum kernel version required has been bumped from 3.13 to 4.15,
6003 and CLOCK_BOOTTIME is now assumed to always exist.
6005 * C11 with GNU extensions (aka "gnu11") is now used to build our
6006 components. Public API headers are still restricted to ISO C89.
6008 * In v250, a systemd-networkd feature that automatically configures
6009 routes to addresses specified in AllowedIPs= was added and enabled by
6010 default. However, this causes network connectivity issues in many
6011 existing setups. Hence, it has been disabled by default since
6012 systemd-stable 250.3. The feature can still be used by explicitly
6013 configuring RouteTable= setting in .netdev files.
6015 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
6016 when a Condition*= check does not succeed, restoring the JobRemoved
6017 signal to the behaviour it had before v250.
6019 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
6020 GetImageMetadataWithExtensions() have been fixed to provide an extra
6021 return parameter, containing the actual extension release metadata.
6022 The current implementation was judged to be broken and unusable, and
6023 thus the usual procedure of adding a new set of methods was skipped,
6024 and backward compatibility broken instead on the assumption that
6025 nobody can be affected given the current state of this interface.
6027 * All kernels supported by systemd mix bytes returned by RDRAND (or
6028 similar) into the entropy pool at early boot. This means that on
6029 those systems, even if /dev/urandom is not yet initialized, it still
6030 returns bytes that are of at least RDRAND quality. For that reason,
6031 we no longer have reason to invoke RDRAND from systemd itself, which
6032 has historically been a source of bugs. Furthermore, kernels ≥5.6
6033 provide the getrandom(GRND_INSECURE) interface for returning random
6034 bytes before the entropy pool is initialized without warning into
6035 kmsg, which is what we attempt to use if available. systemd's direct
6036 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
6037 running an older kernel may experience kmsg warnings that were not
6038 seen with 250. For newer kernels, non-x86 systems, or older x86
6039 systems, there should be no visible changes.
6041 * sd-boot will now measure the kernel command line into TPM PCR 12
6042 rather than PCR 8. This improves usefulness of the measurements on
6043 systems where sd-boot is chainloaded from Grub. Grub measures all
6044 commands its executes into PCR 8, which makes it very hard to use
6045 reasonably, hence separate ourselves from that and use PCR 12
6046 instead, which is what certain Ubuntu editions already do. To retain
6047 compatibility with systems running older systemd systems a new meson
6048 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
6049 If enabled, the measurement is done twice: into the new-style PCR 12
6050 *and* the old-style PCR 8. It's strongly advised to migrate all users
6051 to PCR 12 for this purpose in the long run, as we intend to remove
6052 this compatibility feature in two years' time.
6054 * busctl capture now writes output in the newer pcapng format instead
6057 * A udev rule that imported hwdb matches for USB devices with lowercase
6058 hexadecimal vendor/product ID digits was added in systemd 250. This
6059 has been reverted, since uppercase hexadecimal digits are supposed to
6060 be used, and we already had a rule with the appropriate match.
6062 Users might need to adjust their local hwdb entries.
6064 * arch_prctl(2) has been moved to the @default set in the syscall filters
6065 (as exposed via the SystemCallFilter= setting in service unit files).
6066 It is apparently used by the linker now.
6068 * The tmpfiles entries that create the /run/systemd/netif directory and
6069 its subdirectories were moved from tmpfiles.d/systemd.conf to
6070 tmpfiles.d/systemd-network.conf.
6072 Users might need to adjust their files that override tmpfiles.d/systemd.conf
6073 to account for this change.
6075 * The requirement for Portable Services images to contain a well-formed
6076 os-release file (i.e.: contain at least an ID field) is now enforced.
6077 This applies to base images and extensions, and also to systemd-sysext.
6079 Changes in the Boot Loader Specification, kernel-install and sd-boot:
6081 * kernel-install's and bootctl's Boot Loader Specification Type #1
6082 entry generation logic has been reworked. The user may now pick
6083 explicitly by which "token" string to name the installation's boot
6084 entries, via the new /etc/kernel/entry-token file or the new
6085 --entry-token= switch to bootctl. By default — as before — the
6086 entries are named after the local machine ID. However, in "golden
6087 image" environments, where the machine ID shall be initialized on
6088 first boot (as opposed to at installation time before first boot) the
6089 machine ID will not be available at build time. In this case the
6090 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
6091 file) may be used to override the "token" for the entries, for
6092 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
6093 make the OS images independent of any machine ID, and ensure that the
6094 images will not carry any identifiable information before first boot,
6095 but on the other hand means that multiple parallel installations of
6096 the very same image on the same disk cannot be supported.
6098 Summary: if you are building golden images that shall acquire
6099 identity information exclusively on first boot, make sure to both
6100 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
6101 value of the IMAGE_ID= or ID= field of /etc/os-release or another
6102 suitable identifier before deploying the image.
6104 * The Boot Loader Specification has been extended with
6105 /loader/entries.srel file located in the EFI System Partition (ESP)
6106 that disambiguates the format of the entries in the /loader/entries/
6107 directory (in order to discern them from incompatible uses of this
6108 directory by other projects). For entries that follow the
6109 Specification, the string "type1" is stored in this file.
6111 bootctl will now write this file automatically when installing the
6112 systemd-boot boot loader.
6114 * kernel-install supports a new initrd_generator= setting in
6115 /etc/kernel/install.conf, that is exported as
6116 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
6117 allows choosing different initrd generators.
6119 * kernel-install will now create a "staging area" (an initially-empty
6120 directory to gather files for a Boot Loader Specification Type #1
6121 entry). The path to this directory is exported as
6122 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
6123 drop files there instead of writing them directly to the final
6124 location. kernel-install will move them when all files have been
6125 prepared successfully.
6127 * New option sort-key= has been added to the Boot Loader Specification
6128 to override the sorting order of the entries in the boot menu. It is
6129 read by sd-boot and bootctl, and will be written by kernel-install,
6130 with the default value of IMAGE_ID= or ID= fields from
6131 os-release. Together, this means that on multiboot installations,
6132 entries should be grouped and sorted in a predictable way.
6134 * The sort order of boot entries has been updated: entries which have
6135 the new field sort-key= are sorted by it first, and all entries
6136 without it are ordered later. After that, entries are sorted by
6137 version so that newest entries are towards the beginning of the list.
6139 * The kernel-install tool gained a new 'inspect' verb which shows the
6140 paths and other settings used.
6142 * sd-boot can now optionally beep when the menu is shown and menu
6143 entries are selected, which can be useful on machines without a
6144 working display. (Controllable via a loader.conf setting.)
6146 * The --make-machine-id-directory= switch to bootctl has been replaced
6147 by --make-entry-directory=, given that the entry directory is not
6148 necessarily named after the machine ID, but after some other suitable
6149 ID as selected via --entry-token= described above. The old name of
6150 the option is still understood to maximize compatibility.
6152 * 'bootctl list' gained support for a new --json= switch to output boot
6153 menu entries in JSON format.
6155 * 'bootctl is-installed' now supports the --graceful, and various verbs
6156 omit output with the new option --quiet.
6158 Changes in systemd-homed:
6160 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
6161 of activated home directories it manages (if the kernel and selected
6162 file systems support it). So far it mapped three UID ranges: the
6163 range from 0…60000, the user's own UID, and the range 60514…65534,
6164 leaving everything else unmapped (in other words, the 16-bit UID range
6165 is mapped almost fully, with the exception of the UID subrange used
6166 for systemd-homed users, with one exception: the user's own UID).
6167 Unmapped UIDs may not be used for file ownership in the home
6168 directory — any chown() attempts with them will fail. With this
6169 release a fourth range is added to these mappings:
6170 524288…1879048191. This range is the UID range intended for container
6173 https://systemd.io/UIDS-GIDS
6175 This range may be used for container managers that place container OS
6176 trees in the home directory (which is a questionable approach, for
6177 quota, permission, SUID handling and network file system
6178 compatibility reasons, but nonetheless apparently commonplace). Note
6179 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
6180 UID assignments from the range are not managed or mapped by
6181 systemd-homed, and must be managed with other mechanisms, in the
6182 context of the local system.
6184 Typically, a better approach to user namespacing in relevant
6185 container managers would be to leave container OS trees on disk at
6186 UID offset 0, but then map them to a dynamically allocated runtime
6187 UID range via another UID mount map at container invocation
6188 time. That way user namespace UID ranges become strictly a runtime
6189 concept, and do not leak into persistent file systems, persistent
6190 user databases or persistent configuration, thus greatly simplifying
6191 handling, and improving compatibility with home directories intended
6192 to be portable like the ones managed by systemd-homed.
6194 Changes in shared libraries:
6196 * A new libsystemd-core-<version>.so private shared library is
6197 installed under /usr/lib/systemd/system, mirroring the existing
6198 libsystemd-shared-<version>.so library. This allows the total
6199 installation size to be reduced by binary code reuse.
6201 * The <version> tag used in the name of libsystemd-shared.so and
6202 libsystemd-core.so can be configured via the meson option
6203 'shared-lib-tag'. Distributions may build subsequent versions of the
6204 systemd package with unique tags (e.g. the full package version),
6205 thus allowing multiple installations of those shared libraries to be
6206 available at the same time. This is intended to fix an issue where
6207 programs that link to those libraries would fail to execute because
6208 they were installed earlier or later than the appropriate version of
6211 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
6212 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
6213 format instead of as a simple series of hex characters.
6215 * The sd-device API gained two new calls sd_device_new_from_devname()
6216 and sd_device_new_from_path() which permit allocating an sd_device
6217 object from a device node name or file system path.
6219 * sd-device also gained a new call sd_device_open() which will open the
6220 device node associated with a device for which an sd_device object
6221 has been allocated. The call is supposed to address races around
6222 device nodes being removed/recycled due to hotplug events, or media
6223 change events: the call checks internally whether the major/minor of
6224 the device node and the "diskseq" (in case of block devices) match
6225 with the metadata loaded in the sd_device object, thus ensuring that
6226 the device once opened really matches the provided sd_device object.
6228 Changes in PID1, systemctl, and systemd-oomd:
6230 * A new set of service monitor environment variables will be passed to
6231 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
6232 handler unit as OnFailure=/OnSuccess=. The variables are:
6233 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
6234 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
6235 handler needs to watch multiple units, use a templated handler.
6237 * A new ExtensionDirectories= setting in service unit files allows
6238 system extensions to be loaded from a directory. (It is similar to
6239 ExtensionImages=, but takes paths to directories, instead of
6242 'portablectl attach --extension=' now also accepts directory paths.
6244 * The user.delegate and user.invocation_id extended attributes on
6245 cgroups are used in addition to trusted.delegate and
6246 trusted.invocation_id. The latter pair requires privileges to set,
6247 but the former doesn't and can be also set by the unprivileged user
6250 (Only supported on kernels ≥5.6.)
6252 * Units that were killed by systemd-oomd will now have a service result
6253 of 'oom-kill'. The number of times a service was killed is tallied
6254 in the 'user.oomd_ooms' extended attribute.
6256 The OOMPolicy= unit file setting is now also honoured by
6259 * In unit files the new %y/%Y specifiers can be used to refer to
6260 normalized unit file path, which is particularly useful for symlinked
6263 The new %q specifier resolves to the pretty hostname
6264 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
6266 The new %d specifier resolves to the credentials directory of a
6267 service (same as $CREDENTIALS_DIRECTORY).
6269 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
6270 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
6271 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
6272 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
6273 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
6274 MountFlags= service settings now also work in unprivileged user
6275 services, i.e. those run by the user's --user service manager, as long
6276 as user namespaces are enabled on the system.
6278 * Services with Restart=always and a failing ExecCondition= will no
6279 longer be restarted, to bring ExecCondition= behaviour in line with
6280 Condition*= settings.
6282 * LoadCredential= now accepts a directory as the argument; all files
6283 from the directory will be loaded as credentials.
6285 * A new D-Bus property ControlGroupId is now exposed on service units,
6286 that encapsulates the service's numeric cgroup ID that newer kernels
6287 assign to each cgroup.
6289 * PID 1 gained support for configuring the "pre-timeout" of watchdog
6290 devices and the associated governor, via the new
6291 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
6292 options in /etc/systemd/system.conf.
6294 * systemctl's --timestamp= option gained a new choice "unix", to show
6295 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
6297 * A new "taint" flag named "old-kernel" is introduced which is set when
6298 the kernel systemd runs on is older then the current baseline version
6299 (see above). The flag is shown in "systemctl status" output.
6301 * Two additional taint flags "short-uid-range" and "short-gid-range"
6302 have been added as well, which are set when systemd notices it is run
6303 within a userns namespace that does not define the full 0…65535 UID
6306 * A new "unmerged-usr" taint flag has been added that is set whenever
6307 running on systems where /bin/ + /sbin/ are *not* symlinks to their
6308 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
6311 * Generators invoked by PID 1 will now have a couple of useful
6312 environment variables set describing the execution context a
6313 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
6314 system service manager, or from the per-user service
6315 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
6316 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
6317 systemd considers the current boot to be a "first"
6318 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
6319 detected and which type of hypervisor/container
6320 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
6321 kernel is built for.
6323 * PID 1 will now automatically pick up system credentials from qemu's
6324 fw_cfg interface, thus allowing passing arbitrary data into VM
6325 systems similar to how this is already supported for passing them
6326 into systemd-nspawn containers. Credentials may now also be passed in
6327 via the new kernel command line option "systemd.set_credential="
6328 (note that kernel command line options are world-readable during
6329 runtime, and only useful for credentials that require no
6330 confidentiality). The credentials that can be passed to unified
6331 kernels that use the systemd-stub UEFI stub are now similarly
6332 picked up automatically. Automatic importing of system credentials
6333 this way can be turned off via the new
6334 "systemd.import_credentials=no" kernel command line option.
6336 * LoadCredential= will now automatically look for credentials in the
6337 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
6338 the argument is not an absolute path. Similarly,
6339 LoadCredentialEncrypted= will check the same directories plus
6340 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
6341 /usr/lib/credstore.encrypted/. The idea is to use those directories
6342 as the system-wide location for credentials that services should pick
6345 * System and service credentials are described in great detail in a new
6348 https://systemd.io/CREDENTIALS
6350 Changes in systemd-journald:
6352 * The journal JSON export format has been added to listed of stable
6353 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
6355 * journalctl --list-boots now supports JSON output and the --reverse option.
6357 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
6358 updated, BUILDING_IMAGES is new:
6360 https://systemd.io/JOURNAL_EXPORT_FORMATS
6361 https://systemd.io/BUILDING_IMAGES
6365 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
6366 calculators, etc.), the other AV production devices (DJ tables,
6367 keypads, etc.) that should accessible to the seat owner user by
6370 * udevadm trigger gained a new --prioritized-subsystem= option to
6371 process certain subsystems (and all their parent devices) earlier.
6373 systemd-udev-trigger.service now uses this new option to trigger
6374 block and TPM devices first, hopefully making the boot a bit faster.
6376 * udevadm trigger now implements --type=all, --initialized-match,
6377 --initialized-nomatch to trigger both subsystems and devices, only
6378 already-initialized devices, and only devices which haven't been
6379 initialized yet, respectively.
6381 * udevadm gained a new "wait" command for safely waiting for a specific
6382 device to show up in the udev device database. This is useful in
6383 scripts that asynchronously allocate a block device (e.g. through
6384 repartitioning, or allocating a loopback device or similar) and need
6385 to synchronize on the creation to complete.
6387 * udevadm gained a new "lock" command for locking one or more block
6388 devices while formatting it or writing a partition table to it. It is
6389 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
6390 usable in scripts dealing with block devices.
6392 * udevadm info will show a couple of additional device fields in its
6393 output, and will not apply a limited set of coloring to line types.
6395 * udevadm info --tree will now show a tree of objects (i.e. devices and
6396 suchlike) in the /sys/ hierarchy.
6398 * Block devices will now get a new set of device symlinks in
6399 /dev/disk/by-diskseq/<nr>, which may be used to reference block
6400 device nodes via the kernel's "diskseq" value. Note that this does
6401 not guarantee that opening a device by a symlink like this will
6402 guarantee that the opened device actually matches the specified
6403 diskseq value. To be safe against races, the actual diskseq value of
6404 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
6405 the one in the symlink path.
6407 * .link files gained support for setting MDI/MID-X on a link.
6409 * .link files gained support for [Match] Firmware= setting to match on
6410 the device firmware description string. By mistake, it was previously
6411 only supported in .network files.
6413 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
6414 and [SR-IOV] section to configure SR-IOV virtual functions.
6416 Changes in systemd-networkd:
6418 * The default scope for unicast routes configured through [Route]
6419 section is changed to "link", to make the behavior consistent with
6420 "ip route" command. The manual configuration of [Route] Scope= is
6423 * A new unit systemd-networkd-wait-online@<interface>.service has been
6424 added that can be used to wait for a specific network interface to be
6427 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
6428 that configures the eponymous kernel attribute on the bridge.
6430 * .netdev files now can be used to create virtual WLAN devices, and
6431 configure various settings on them, via the [WLAN] section.
6433 * .link/.network files gained support for [Match] Kind= setting to match
6434 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
6436 This value is also shown by 'networkctl status'.
6438 * The Local= setting in .netdev files for various virtual network
6439 devices gained support for specifying, in addition to the network
6440 address, the name of a local interface which must have the specified
6443 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
6444 files, to configure tunnels in external mode (a.k.a. collect metadata
6447 * [Network] L2TP= setting was removed. Please use interface specifier in
6448 Local= setting in .netdev files of corresponding L2TP interface.
6450 * New [DHCPServer] BootServerName=, BootServerAddress=, and
6451 BootFilename= settings can be used to configure the server address,
6452 server name, and file name sent in the DHCP packet (e.g. to configure
6455 Changes in systemd-resolved:
6457 * systemd-resolved is started earlier (in sysinit.target), so it
6458 available earlier and will also be started in the initrd if installed
6461 Changes in disk encryption:
6463 * systemd-cryptenroll can now control whether to require the user to
6464 enter a PIN when using TPM-based unlocking of a volume via the new
6465 --tpm2-with-pin= option.
6467 Option tpm2-pin= can be used in /etc/crypttab.
6469 * When unlocking devices via TPM, TPM2 parameter encryption is now
6470 used, to ensure that communication between CPU and discrete TPM chips
6471 cannot be eavesdropped to acquire disk encryption keys.
6473 * A new switch --fido2-credential-algorithm= has been added to
6474 systemd-cryptenroll allowing selection of the credential algorithm to
6475 use when binding encryption to FIDO2 tokens.
6477 Changes in systemd-hostnamed:
6479 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
6480 to override the values gleaned from the hwdb.
6482 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
6483 /sys/class/dmi/id) to override the chassis that is reported by
6486 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
6487 for reading the hardware serial number, as reported by DMI. It also
6488 exposes a new method D-Bus property FirmwareVersion that encode the
6489 firmware version of the system.
6491 Changes in other components:
6493 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
6494 handling with the values that were configured during systemd build
6495 (if /etc/locale.conf has not been created through some other
6496 mechanism). This means that /etc/locale.conf should always have
6497 reasonable contents and we avoid a potential mismatch in defaults.
6499 * The userdbctl tool will now show UID range information as part of the
6500 list of known users.
6502 * A new build-time configuration setting default-user-shell= can be
6503 used to set the default shell for user records and nspawn shell
6504 invocations (instead of the default /bin/bash).
6506 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
6507 information dynamically at runtime via IPC.
6509 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
6510 whether a functioning TPM2 infrastructure is available, i.e. if
6511 firmware, kernel driver and systemd all have TPM2 support enabled and
6514 * The systemd-creds tool gained support for generating encrypted
6515 credentials that are using an empty encryption key. While this
6516 provides no integrity nor confidentiality it's useful to implement
6517 codeflows that work the same on TPM-ful and TPM2-less systems. The
6518 service manager will only accept credentials "encrypted" that way if
6519 a TPM2 device cannot be detected, to ensure that credentials
6520 "encrypted" like that cannot be used to trick TPM2 systems.
6522 * When deciding whether to colorize output, all systemd programs now
6523 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
6526 * Meson's new install_tag feature is now in use for several components,
6527 allowing to build and install select binaries only: pam, nss, devel
6528 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
6529 $ meson build systemd-boot
6530 $ meson install --tags systemd-boot --no-rebuild
6531 https://mesonbuild.com/Installing.html#installation-tags
6533 * A new build configuration option has been added, to allow selecting the
6534 default compression algorithm used by systemd-journald and systemd-coredump.
6535 This allows to build-in support for decompressing all supported formats,
6536 but choose a specific one for compression. E.g.:
6537 $ meson -Ddefault-compression=xz
6539 Experimental features:
6541 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
6542 loader.conf that implements booting Microsoft Windows from the
6543 sd-boot in a way that first reboots the system, to reset the TPM
6544 PCRs. This improves compatibility with BitLocker's TPM use, as the
6545 PCRs will only record the Windows boot process, and not sd-boot
6546 itself, thus retaining the PCR measurements not involving sd-boot.
6547 Note that this feature is experimental for now, and is likely going
6548 to be generalized and renamed in a future release, without retaining
6549 compatibility with the current implementation.
6551 * A new systemd-sysupdate component has been added that automatically
6552 discovers, downloads, and installs A/B-style updates for the host
6553 installation itself, or container images, portable service images,
6554 and other assets. See the new systemd-sysupdate man page for updates.
6556 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
6557 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
6558 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
6559 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
6560 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
6561 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
6562 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
6563 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
6564 David, David Bond, Davide Cavalca, David Tardon, davijosw,
6565 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
6566 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
6567 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
6568 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
6569 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
6570 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
6571 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
6572 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
6573 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
6574 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
6575 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
6576 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
6577 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
6578 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
6579 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
6580 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
6581 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
6582 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
6583 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
6584 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
6585 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
6586 Simon Ellmann, Sonali Srivastava, Stefan Seering,
6587 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
6588 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
6589 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
6590 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
6591 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
6592 Zbigniew Jędrzejewski-Szmek, наб
6594 — Edinburgh, 2022-05-21
6598 * Support for encrypted and authenticated credentials has been added.
6599 This extends the credential logic introduced with v247 to support
6600 non-interactive symmetric encryption and authentication, based on a
6601 key that is stored on the /var/ file system or in the TPM2 chip (if
6602 available), or the combination of both (by default if a TPM2 chip
6603 exists the combination is used, otherwise the /var/ key only). The
6604 credentials are automatically decrypted at the moment a service is
6605 started, and are made accessible to the service itself in unencrypted
6606 form. A new tool 'systemd-creds' encrypts credentials for this
6607 purpose, and two new service file settings LoadCredentialEncrypted=
6608 and SetCredentialEncrypted= configure such credentials.
6610 This feature is useful to store sensitive material such as SSL
6611 certificates, passwords and similar securely at rest and only decrypt
6612 them when needed, and in a way that is tied to the local OS
6613 installation or hardware.
6615 * systemd-gpt-auto-generator can now automatically set up discoverable
6616 LUKS2 encrypted swap partitions.
6618 * The GPT Discoverable Partitions Specification has been substantially
6619 extended with support for root and /usr/ partitions for the majority
6620 of architectures systemd supports. This includes platforms that do
6621 not natively support UEFI, because even though GPT is specified under
6622 UEFI umbrella, it is useful on other systems too. Specifically,
6623 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
6624 Portable Services use the concept without requiring UEFI.
6626 * The GPT Discoverable Partitions Specifications has been extended with
6627 a new set of partitions that may carry PKCS#7 signatures for Verity
6628 partitions, encoded in a simple JSON format. This implements a simple
6629 mechanism for building disk images that are fully authenticated and
6630 can be tested against a set of cryptographic certificates. This is
6631 now implemented for the various systemd tools that can operate with
6632 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
6633 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
6634 The PKCS#7 signatures are passed to the kernel (where they are
6635 checked against certificates from the kernel keyring), or can be
6636 verified against certificates provided in userspace (via a simple
6637 drop-in file mechanism).
6639 * systemd-dissect's inspection logic will now report for which uses a
6640 disk image is intended. Specifically, it will display whether an
6641 image is suitable for booting on UEFI or in a container (using
6642 systemd-nspawn's --image= switch), whether it can be used as portable
6643 service, or attached as system extension.
6645 * The system-extension.d/ drop-in files now support a new field
6646 SYSEXT_SCOPE= that may encode which purpose a system extension image
6647 is for: one of "initrd", "system" or "portable". This is useful to
6648 make images more self-descriptive, and to ensure system extensions
6649 cannot be attached in the wrong contexts.
6651 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
6652 be used in portable service images to indicate which unit prefixes
6655 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
6656 now is able to decode images for non-native architectures as well.
6657 This allows systemd-nspawn to boot images of non-native architectures
6658 if the corresponding user mode emulator is installed and
6659 systemd-binfmtd is running.
6661 * systemd-logind gained new settings HandlePowerKeyLongPress=,
6662 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
6663 HandleHibernateKeyLongPress= which may be used to configure actions
6664 when the relevant keys are pressed for more than 5s. This is useful
6665 on devices that only have hardware for a subset of these keys. By
6666 default, if the reboot key is pressed long the poweroff operation is
6667 now triggered, and when the suspend key is pressed long the hibernate
6668 operation is triggered. Long pressing the other two keys currently
6669 does not trigger any operation by default.
6671 * When showing unit status updates on the console during boot and
6672 shutdown, and a service is slow to start so that the cylon animation
6673 is shown, the most recent sd_notify() STATUS= text is now shown as
6674 well. Services may use this to make the boot/shutdown output easier
6675 to understand, and to indicate what precisely a service that is slow
6676 to start or stop is waiting for. In particular, the per-user service
6677 manager instance now reports what it is doing and which service it is
6678 waiting for this way to the system service manager.
6680 * The service manager will now re-execute on reception of the
6681 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
6682 only when running as PID 1. There was no signal to request this when
6683 running as per-user service manager, i.e. as any other PID than 1.
6684 SIGRTMIN+25 works for both system and user managers.
6686 * The hardware watchdog logic in PID 1 gained support for operating
6687 with the default timeout configured in the hardware, instead of
6688 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
6689 request this behavior.
6691 * A new kernel command line option systemd.watchdog_sec= is now
6692 understood which may be used to override the hardware watchdog
6693 timeout for the boot.
6695 * A new setting DefaultOOMScoreAdjust= is now supported in
6696 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
6697 to set the default process OOM score adjustment value for processes
6698 started by the service manager. For per-user service managers this
6699 now defaults to 100, but for per-system service managers is left as
6700 is. This means that by default now services forked off the user
6701 service manager are more likely to be killed by the OOM killer than
6702 system services or the managers themselves.
6704 * A new per-service setting RestrictFileSystems= as been added that
6705 restricts the file systems a service has access to by their type.
6706 This is based on the new BPF LSM of the Linux kernel. It provides an
6707 effective way to make certain API file systems unavailable to
6708 services (and thus minimizing attack surface). A new command
6709 "systemd-analyze filesystems" has been added that lists all known
6710 file system types (and how they are grouped together under useful
6713 * Services now support a new setting RestrictNetworkInterfaces= for
6714 restricting access to specific network interfaces.
6716 * Service unit files gained new settings StartupAllowedCPUs= and
6717 StartupAllowedMemoryNodes=. These are similar to their counterparts
6718 without the "Startup" prefix and apply during the boot process
6719 only. This is useful to improve boot-time behavior of the system and
6720 assign resources differently during boot than during regular
6721 runtime. This is similar to the preexisting StartupCPUWeight=
6724 * Related to this: the various StartupXYZ= settings
6725 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
6726 during shutdown. The settings not prefixed with "Startup" hence apply
6727 during regular runtime, and those that are prefixed like that apply
6728 during boot and shutdown.
6730 * A new per-unit set of conditions/asserts
6731 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
6732 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
6733 pressure is above the configured threshold, using the kernel PSI
6734 feature. For more details see systemd.unit(5) and
6735 https://docs.kernel.org/accounting/psi.html
6737 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
6738 ProtectKernelLogs=yes can now be used.
6740 * The default maximum numbers of inodes have been raised from 64k to 1M
6741 for /dev/, and from 400k to 1M for /tmp/.
6743 * The per-user service manager learnt support for communicating with
6744 systemd-oomd to acquire OOM kill information.
6746 * A new service setting ExecSearchPath= has been added that allows
6747 changing the search path for executables for services. It affects
6748 where we look for the binaries specified in ExecStart= and similar,
6749 and the specified directories are also added the $PATH environment
6750 variable passed to invoked processes.
6752 * A new setting RuntimeRandomizedExtraSec= has been added for service
6753 and scope units that allows extending the runtime timeout as
6754 configured by RuntimeMaxSec= with a randomized amount.
6756 * The syntax of the service unit settings RuntimeDirectory=,
6757 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
6758 if the specified value is now suffixed with a colon, followed by
6759 another filename, the latter will be created as symbolic link to the
6760 specified directory. This allows creating these service directories
6761 together with alias symlinks to make them available under multiple
6764 * Service unit files gained two new settings TTYRows=/TTYColumns= for
6765 configuring rows/columns of the TTY device passed to
6766 stdin/stdout/stderr of the service. This is useful to propagate TTY
6767 dimensions to a virtual machine.
6769 * A new service unit file setting ExitType= has been added that
6770 specifies when to assume a service has exited. By default systemd
6771 only watches the main process of a service. By setting
6772 ExitType=cgroup it can be told to wait for the last process in a
6775 * Automount unit files gained a new setting ExtraOptions= that can be
6776 used to configure additional mount options to pass to the kernel when
6777 mounting the autofs instance.
6779 * "Urlification" (generation of ESC sequences that generate clickable
6780 hyperlinks in modern terminals) may now be turned off altogether
6783 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
6784 settings that default to 200 and 2 s respectively. The ratelimit
6785 ensures that a path unit cannot cause PID1 to busy-loop when it is
6786 trying to trigger a service that is skipped because of a Condition*=
6787 not being satisfied. This matches the configuration and behaviour of
6790 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
6791 as a plug-in for cryptsetup. This means the plain cryptsetup command
6792 may now be used to unlock volumes set up this way.
6794 * The TPM2 logic in cryptsetup will now automatically detect systems
6795 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
6796 updates the SHA1 banks. In such a case PCR policies will be
6797 automatically bound to the latter, not the former. This makes the PCR
6798 policies reliable, but of course do not provide the same level of
6799 trust as SHA256 banks.
6801 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
6802 RSA primary keys in addition to ECC, improving compatibility with
6803 TPM2 chips that do not support ECC. RSA keys are much slower to use
6804 than ECC, and hence are only used if ECC is not available.
6806 * /etc/crypttab gained support for a new token-timeout= setting for
6807 encrypted volumes that allows configuration of the maximum time to
6808 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
6809 the logic will query the user for a regular passphrase/recovery key
6812 * Support for activating dm-integrity volumes at boot via a new file
6813 /etc/integritytab and the tool systemd-integritysetup have been
6814 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
6815 with dm-integrity instead of dm-crypt/dm-verity.
6817 * The systemd-veritysetup-generator now understands a new usrhash=
6818 kernel command line option for specifying the Verity root hash for
6819 the partition backing the /usr/ file system. A matching set of
6820 systemd.verity_usr_* kernel command line options has been added as
6821 well. These all work similar to the corresponding options for the
6824 * The sd-device API gained a new API call sd_device_get_diskseq() to
6825 return the DISKSEQ property of a device structure. The "disk
6826 sequence" concept is a new feature recently introduced to the Linux
6827 kernel that allows detecting reuse cycles of block devices, i.e. can
6828 be used to recognize when loopback block devices are reused for a
6829 different purpose or CD-ROM drives get their media changed.
6831 * A new unit systemd-boot-update.service has been added. If enabled
6832 (the default) and the sd-boot loader is detected to be installed, it
6833 is automatically updated to the newest version when out of date. This
6834 is useful to ensure the boot loader remains up-to-date, and updates
6835 automatically propagate from the OS tree in /usr/.
6837 * sd-boot will now build with SBAT by default in order to facilitate
6838 working with recent versions of Shim that require it to be present.
6840 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
6841 This is used to robustly generate boot entry titles for Windows.
6843 * A new generic target unit factory-reset.target has been added. It is
6844 hooked into systemd-logind similar in fashion to
6845 reboot/poweroff/suspend/hibernate, and is supposed to be used to
6846 initiate a factory reset operation. What precisely this operation
6847 entails is up for the implementer to decide, the primary goal of the
6848 new unit is provide a framework where to plug in the implementation
6849 and how to trigger it.
6851 * A new meson build-time option 'clock-valid-range-usec-max' has been
6852 added which takes a time in µs and defaults to 15 years. If the RTC
6853 time is noticed to be more than the specified time ahead of the
6854 built-in epoch of systemd (which by default is the release timestamp
6855 of systemd) it is assumed that the RTC is not working correctly, and
6856 the RTC is reset to the epoch. (It already is reset to the epoch when
6857 noticed to be before it.) This should increase the chance that time
6858 doesn't accidentally jump too far ahead due to faulty hardware or
6861 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
6862 which may be used to automatically save the current system time to
6863 disk in regular intervals. This is useful to maintain a roughly
6864 monotonic clock even without RTC hardware and with some robustness
6865 against abnormal system shutdown.
6867 * systemd-analyze verify gained support for a pair of new --image= +
6868 --root= switches for verifying units below a specific root
6869 directory/image instead of on the host.
6871 * systemd-analyze verify gained support for verifying unit files under
6872 an explicitly specified unit name, independently of what the filename
6875 * systemd-analyze verify gained a new switch --recursive-errors= which
6876 controls whether to only fail on errors found in the specified units
6877 or recursively any dependent units.
6879 * systemd-analyze security now supports a new --offline mode for
6880 analyzing unit files stored on disk instead of loaded units. It may
6881 be combined with --root=/--image to analyze unit files under a root
6882 directory or disk image. It also learnt a new --threshold= parameter
6883 for specifying an exposure level threshold: if the exposure level
6884 exceeds the specified value the call will fail. It also gained a new
6885 --security-policy= switch for configuring security policies to
6886 enforce on the units. A policy is a JSON file that lists which tests
6887 shall be weighted how much to determine the overall exposure
6888 level. Altogether these new features are useful for fully automatic
6889 analysis and enforcement of security policies on unit files.
6891 * systemd-analyze security gain a new --json= switch for JSON output.
6893 * systemd-analyze learnt a new --quiet switch for reducing
6894 non-essential output. It's honored by the "dot", "syscall-filter",
6895 "filesystems" commands.
6897 * systemd-analyze security gained a --profile= option that can be used
6898 to take into account a portable profile when analyzing portable
6899 services, since a lot of the security-related settings are enabled
6902 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
6903 files, binaries and executables and prints metadata information,
6904 including the build-id and other info described on:
6905 https://systemd.io/COREDUMP_PACKAGE_METADATA/
6907 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
6908 section, for automatically propagating DNS settings from other
6911 * The static lease DHCP server logic in systemd-networkd may now serve
6912 IP addresses outside of the configured IP pool range for the server.
6914 * CAN support in systemd-networkd gained four new settings Loopback=,
6915 OneShot=, PresumeACK=, ClassicDataLengthCode= for tweaking CAN
6916 control modes. It gained a number of further settings for tweaking
6919 * The [CAN] section in .network file gained new TimeQuantaNSec=,
6920 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
6921 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
6922 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
6923 DataSyncJumpWidth= settings to control bit-timing processed by the
6926 * DHCPv4 client support in systemd-networkd learnt a new Label= option
6927 for configuring the address label to apply to configure IPv4
6930 * The [IPv6AcceptRA] section of .network files gained support for a new
6931 UseMTU= setting that may be used to control whether to apply the
6932 announced MTU settings to the local interface.
6934 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
6935 setting to control whether the DHCPv4 client request and process the
6938 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
6939 [DHCPPrefixDelegation], as now the prefix delegation is also supported
6940 with DHCPv4 protocol by enabling the Use6RD= setting.
6942 * The [DHCPPrefixDelegation] section in .network file gained a new
6943 setting UplinkInterface= to specify the upstream interface.
6945 * The [DHCPv6] section in .network file gained a new setting
6946 UseDelegatedPrefix= to control whether the delegated prefixes will be
6947 propagated to the downstream interfaces.
6949 * The [IPv6AcceptRA] section of .network files now understands two new
6950 settings UseGateway=/UseRoutePrefix= for explicitly configuring
6951 whether to use the relevant fields from the IPv6 Router Advertisement
6954 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
6955 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
6956 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
6957 [IPv6AcceptRA] section to control when the DHCPv6 client is started
6958 and how the delegated prefixes are handled by the DHCPv6 client.
6960 * The IPv6Token= section in the [Network] section is deprecated, and
6961 the [IPv6AcceptRA] section gained the Token= setting for its
6962 replacement. The [IPv6Prefix] section also gained the Token= setting.
6963 The Token= setting gained 'eui64' mode to explicitly configure an
6964 address with the EUI64 algorithm based on the interface MAC address.
6965 The 'prefixstable' mode can now optionally take a secret key. The
6966 Token= setting in the [DHCPPrefixDelegation] section now supports all
6967 algorithms supported by the same settings in the other sections.
6969 * The [RoutingPolicyRule] section of .network file gained a new
6970 SuppressInterfaceGroup= setting.
6972 * The IgnoreCarrierLoss= setting in the [Network] section of .network
6973 files now allows a duration to be specified, controlling how long to
6974 wait before reacting to carrier loss.
6976 * The [DHCPServer] section of .network file gained a new Router=
6977 setting to specify the router address.
6979 * The [CAKE] section of .network files gained various new settings
6980 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
6981 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
6982 and UseRawPacketSize= for configuring CAKE.
6984 * systemd-networkd now ships with new default .network files:
6985 80-container-vb.network which matches host-side network bridge device
6986 created by systemd-nspawn's --network-bridge or --network-zone
6987 switch, and 80-6rd-tunnel.network which matches automatically created
6988 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
6990 * systemd-networkd's handling of Endpoint= resolution for WireGuard
6991 interfaces has been improved.
6993 * systemd-networkd will now automatically configure routes to addresses
6994 specified in AllowedIPs=. This feature can be controlled via
6995 RouteTable= and RouteMetric= settings in [WireGuard] or
6996 [WireGuardPeer] sections.
6998 * systemd-networkd will now once again automatically generate persistent
6999 MAC addresses for batadv and bridge interfaces. Users can disable this
7000 by using MACAddress=none in .netdev files.
7002 * systemd-networkd and systemd-udevd now support IP over InfiniBand
7003 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
7004 systemd.netdev files gained the [IPoIB] section.
7006 * systemd-networkd and systemd-udevd now support net.ifname_policy=
7007 option on the kernel command-line. This is implemented through the
7008 systemd-network-generator service that automatically generates
7009 appropriate .link, .network, and .netdev files.
7011 * The various systemd-udevd "ethtool" buffer settings now understand
7012 the special value "max" to configure the buffers to the maximum the
7015 * systemd-udevd's .link files may now configure a large variety of
7016 NIC coalescing settings, plus more hardware offload settings.
7018 * .link files gained a new WakeOnLanPassword= setting in the [Link]
7019 section that allows to specify a WoL "SecureOn" password on hardware
7022 * systemd-nspawn's --setenv= switch now supports an additional syntax:
7023 if only a variable name is specified (i.e. without being suffixed by
7024 a '=' character and a value) the current value of the environment
7025 variable is propagated to the container. e.g. --setenv=FOO will
7026 lookup the current value of $FOO in the environment, and pass it down
7027 to the container. Similar behavior has been added to homectl's,
7028 machinectl's and systemd-run's --setenv= switch.
7030 * systemd-nspawn gained a new switch --suppress-sync= which may be used
7031 to optionally suppress the effect of the sync()/fsync()/fdatasync()
7032 system calls for the container payload. This is useful for build
7033 system environments where safety against abnormal system shutdown is
7034 not essential as all build artifacts can be regenerated any time, but
7035 the performance win is beneficial.
7037 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
7038 same value that PID 1 uses for most forked off processes.
7040 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
7041 uidmap/nouidmap options as last parameter. If "uidmap" is used the
7042 bind mounts are created with UID mapping taking place that ensures
7043 the host's file ownerships are mapped 1:1 to container file
7044 ownerships, even if user namespacing is used. This way
7045 files/directories bound into containers will no longer show up as
7046 owned by the nobody user as they typically did if no special care was
7047 taken to shift them manually.
7049 * When discovering Windows installations sd-boot will now attempt to
7050 show the Windows version.
7052 * The color scheme to use in sd-boot may now be configured at
7055 * sd-boot gained the ability to change screen resolution during
7056 boot-time, by hitting the "r" key. This will cycle through available
7057 resolutions and save the last selection.
7059 * sd-boot learnt a new hotkey "f". When pressed the system will enter
7060 firmware setup. This is useful in environments where it is difficult
7061 to hit the right keys early enough to enter the firmware, and works
7062 on any firmware regardless which key it natively uses.
7064 * sd-boot gained support for automatically booting into the menu item
7065 selected on the last boot (using the "@saved" identifier for menu
7068 * sd-boot gained support for automatically loading all EFI drivers
7069 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
7070 Partition (ESP). These drivers are loaded before the menu entries are
7071 loaded. This is useful e.g. to load additional file system drivers
7072 for the XBOOTLDR partition.
7074 * systemd-boot will now paint the input cursor on its own instead of
7075 relying on the firmware to do so, increasing compatibility with broken
7076 firmware that doesn't make the cursor reasonably visible.
7078 * sd-boot now embeds a .osrel PE section like we expect from Boot
7079 Loader Specification Type #2 Unified Kernels. This means sd-boot
7080 itself may be used in place of a Type #2 Unified Kernel. This is
7081 useful for debugging purposes as it allows chain-loading one a
7082 (development) sd-boot instance from another.
7084 * sd-boot now supports a new "devicetree" field in Boot Loader
7085 Specification Type #1 entries: if configured the specified device
7086 tree file is installed before the kernel is invoked. This is useful
7087 for installing/applying new devicetree files without updating the
7090 * Similarly, sd-stub now can read devicetree data from a PE section
7091 ".dtb" and apply it before invoking the kernel.
7093 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
7094 gained the ability to pick up credentials and sysext files, wrap them
7095 in a cpio archive, and pass as an additional initrd to the invoked
7096 Linux kernel, in effect placing those files in the /.extra/ directory
7097 of the initrd environment. This is useful to implement trusted initrd
7098 environments which are fully authenticated but still can be extended
7099 (via sysexts) and parameterized (via encrypted/authenticated
7100 credentials, see above).
7102 Credentials can be located next to the kernel image file (credentials
7103 specific to a single boot entry), or in one of the shared directories
7104 (credentials applicable to multiple boot entries).
7106 * sd-stub now comes with a full man page, that explains its feature set
7107 and how to combine a kernel image, an initrd and the stub to build a
7108 complete EFI unified kernel image, implementing Boot Loader
7109 Specification Type #2.
7111 * sd-stub may now provide the initrd to the executed kernel via the
7112 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
7113 non-x86 architectures.
7115 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
7116 may be used to set the boot menu timeout of the boot loader (for all
7117 or just the subsequent boot).
7119 * bootctl and kernel-install will now read variables
7120 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
7121 /etc/kernel/install.conf. When set, it specifies the layout to use
7122 for installation directories on the boot partition, so that tools
7123 don't need to guess it based on the already-existing directories. The
7124 only value that is defined natively is "bls", corresponding to the
7126 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
7127 kernel-install that implement a different layout can declare other
7128 values for this variable.
7130 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
7131 assumption that if the user installed sd-boot to the ESP, they intend
7132 to use the entry layout understood by sd-boot. It'll also write
7133 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
7134 (and it wasn't specified in the config file yet). Similarly,
7135 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
7136 wasn't specified in the config file yet). Effectively, those changes
7137 mean that the machine-id used for boot loader entry installation is
7138 "frozen" upon first use and becomes independent of the actual
7141 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
7142 images created for distribution ("golden images") are built with no
7143 machine-id, so that a unique machine-id can be created on the first
7144 boot. But those images may contain boot loader entries with the
7145 machine-id used during build included in paths. Using a "frozen"
7146 value allows unambiguously identifying entries that match the
7147 specific installation, while still permitting parallel installations
7150 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
7151 kernel-install to guess the installation layout. This fixes the
7152 problem where a (possibly empty) directory in the boot partition is
7153 created from a different layout causing kernel-install plugins to
7154 assume the wrong layout. A particular example of how this may happen
7155 is the grub2 package in Fedora which includes directories under /boot
7156 directly in its file list. Various other packages pull in grub2 as a
7157 dependency, so it may be installed even if unused, breaking
7158 installations that use the bls layout.
7160 * bootctl and systemd-bless-boot can now be linked statically.
7162 * systemd-sysext now optionally doesn't insist on extension-release.d/
7163 files being placed in the image under the image's file name. If the
7164 file system xattr user.extension-release.strict is set on the
7165 extension release file, it is accepted regardless of its name. This
7166 relaxes security restrictions a bit, as system extension may be
7167 attached under a wrong name this way.
7169 * udevadm's test-builtin command learnt a new --action= switch for
7170 testing the built-in with the specified action (in place of the
7173 * udevadm info gained new switches --property=/--value for showing only
7174 specific udev properties/values instead of all.
7176 * A new hwdb database has been added that contains matches for various
7177 types of signal analyzers (protocol analyzers, logic analyzers,
7178 oscilloscopes, multimeters, bench power supplies, etc.) that should
7179 be accessible to regular users.
7181 * A new hwdb database entry has been added that carries information
7182 about types of cameras (regular or infrared), and in which direction
7183 they point (front or back).
7185 * A new rule to allow console users access to rfkill by default has been
7188 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
7189 now also owned by the system group "sgx".
7191 * A new build-time meson option "extra-net-naming-schemes=" has been
7192 added to define additional naming schemes for udev's network
7193 interface naming logic. This is useful for enterprise distributions
7194 and similar which want to pin the schemes of certain distribution
7195 releases under a specific name and previously had to patch the
7196 sources to introduce new named schemes.
7198 * The predictable naming logic for network interfaces has been extended
7199 to generate stable names from Xen netfront device information.
7201 * hostnamed's chassis property can now be sourced from chassis-type
7202 field encoded in devicetree (in addition to the existing DMI
7205 * systemd-cgls now optionally displays cgroup IDs and extended
7206 attributes for each cgroup. (Controllable via the new --xattr= +
7207 --cgroup-id= switches.)
7209 * coredumpctl gained a new --all switch for operating on all
7210 Journal files instead of just the local ones.
7212 * systemd-coredump will now use libdw/libelf via dlopen() rather than
7213 directly linking, allowing users to easily opt-out of backtrace/metadata
7214 analysis of core files, and reduce image sizes when this is not needed.
7216 * systemd-coredump will now analyze core files with libdw/libelf in a
7217 forked, sandboxed process.
7219 * systemd-homed will now try to unmount an activate home area in
7220 regular intervals once the user logged out fully. Previously this was
7221 attempted exactly once but if the home directory was busy for some
7222 reason it was not tried again.
7224 * systemd-homed's LUKS2 home area backend will now create a BSD file
7225 system lock on the image file while the home area is active
7226 (i.e. mounted). If a home area is found to be locked, logins are
7227 politely refused. This should improve behavior when using home areas
7228 images that are accessible via the network from multiple clients, and
7229 reduce the chance of accidental file system corruption in that case.
7231 * Optionally, systemd-homed will now drop the kernel buffer cache once
7232 a user has fully logged out, configurable via the new --drop-caches=
7235 * systemd-homed now makes use of UID mapped mounts for the home areas.
7236 If the kernel and used file system support it, files are now
7237 internally owned by the "nobody" user (i.e. the user typically used
7238 for indicating "this ownership is not mapped"), and dynamically
7239 mapped to the UID used locally on the system via the UID mapping
7240 mount logic of recent kernels. This makes migrating home areas
7241 between different systems cheaper because recursively chown()ing file
7242 system trees is no longer necessary.
7244 * systemd-homed's CIFS backend now optionally supports CIFS service
7245 names with a directory suffix, in order to place home directories in
7246 a subdirectory of a CIFS share, instead of the top-level directory.
7248 * systemd-homed's CIFS backend gained support for specifying additional
7249 mount options in the JSON user record (cifsExtraMountOptions field,
7250 and --cifs-extra-mount-options= homectl switch). This is for example
7251 useful for configuring mount options such as "noserverino" that some
7252 SMB3 services require (use that to run a homed home directory from a
7253 FritzBox SMB3 share this way).
7255 * systemd-homed will now default to btrfs' zstd compression for home
7256 areas. This is inspired by Fedora's recent decision to switch to zstd
7259 * Additional mount options to use when mounting the file system of
7260 LUKS2 volumes in systemd-homed has been added. Via the
7261 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
7262 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
7263 systemd-homed or via the luksExtraMountOptions user record JSON
7264 property. (Exposed via homectl --luks-extra-mount-options)
7266 * homectl's resize command now takes the special size specifications
7267 "min" and "max" to shrink/grow the home area to the minimum/maximum
7268 size possible, taking disk usage/space constraints and file system
7269 limitations into account. Resizing is now generally graceful: the
7270 logic will try to get as close to the specified size as possible, but
7271 not consider it a failure if the request couldn't be fulfilled
7274 * systemd-homed gained the ability to automatically shrink home areas
7275 on logout to their minimal size and grow them again on next
7276 login. This ensures that while inactive, a home area only takes up
7277 the minimal space necessary, but once activated, it provides
7278 sufficient space for the user's needs. This behavior is only
7279 supported if btrfs is used as file system inside the home area
7280 (because only for btrfs online growing/shrinking is implemented in
7281 the kernel). This behavior is now enabled by default, but may be
7282 controlled via the new --auto-resize-mode= setting of homectl.
7284 * systemd-homed gained support for automatically re-balancing free disk
7285 space among active home areas, in case the LUKS2 backends are used,
7286 and no explicit disk size was requested. This way disk space is
7287 automatically managed and home areas resized in regular intervals and
7288 manual resizing when disk space becomes scarce should not be
7289 necessary anymore. This behavior is only supported if btrfs is used
7290 within the home areas (as only then online shrinking and growing is
7291 supported), and may be configured via the new rebalanceWeight JSON
7292 user record field (as exposed via the new --rebalance-weight= homectl
7293 setting). Re-balancing is mostly automatic, but can also be requested
7294 explicitly via "homectl rebalance", which is synchronous, and thus
7295 may be used to wait until the rebalance run is complete.
7297 * userdbctl gained a --json= switch for configured the JSON formatting
7298 to use when outputting user or group records.
7300 * userdbctl gained a new --multiplexer= switch for explicitly
7301 configuring whether to use the systemd-userdbd server side user
7302 record resolution logic.
7304 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
7305 for chaining up another command to execute after completing the
7306 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
7307 configuration of a single command to invoke, this maybe used to
7308 invoke multiple: first userdbctl's own implementation, and then any
7309 other also configured in the command line.
7311 * The sd-event API gained a new function sd_event_add_inotify_fd() that
7312 is similar to sd_event_add_inotify() but accepts a file descriptor
7313 instead of a path in the file system for referencing the inode to
7316 * The sd-event API gained a new function
7317 sd_event_source_set_ratelimit_expire_callback() that may be used to
7318 define a callback function that is called whenever an event source
7319 leaves the rate limiting phase.
7321 * New documentation has been added explaining which steps are necessary
7322 to port systemd to a new architecture:
7324 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
7326 * The x-systemd.makefs option in /etc/fstab now explicitly supports
7327 ext2, ext3, and f2fs file systems.
7329 * Mount units and units generated from /etc/fstab entries with 'noauto'
7330 are now ordered the same as other units. Effectively, they will be
7331 started earlier (if something actually pulled them in) and stopped
7332 later, similarly to normal mount units that are part of
7333 fs-local.target. This change should be invisible to users, but
7334 should prevent those units from being stopped too early during
7337 * The systemd-getty-generator now honors a new kernel command line
7338 argument systemd.getty_auto= and a new environment variable
7339 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
7340 example useful to turn off gettys inside of containers or similar
7343 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
7344 (in addition to 127.0.0.53, as before). If DNS requests are sent to
7345 this address they are propagated in "bypass" mode only, i.e. are
7346 almost not processed locally, but mostly forwarded as-is to the
7347 current upstream DNS servers. This provides a stable DNS server
7348 address that proxies all requests dynamically to the right upstream
7349 DNS servers even if these dynamically change. This stub does not do
7350 mDNS/LLMNR resolution. However, it will translate look-ups to
7351 DNS-over-TLS if necessary. This new stub is particularly useful in
7352 container/VM environments, or for tethering setups: use DNAT to
7353 redirect traffic to any IP address to this stub.
7355 * systemd-importd now honors new environment variables
7356 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
7357 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
7358 generation, btrfs quota setup and disk synchronization.
7360 * systemd-importd and systemd-resolved can now be optionally built with
7361 OpenSSL instead of libgcrypt.
7363 * systemd-repart no longer requires OpenSSL.
7365 * systemd-sysusers will no longer create the redundant 'nobody' group
7366 by default, as the 'nobody' user is already created with an
7367 appropriate primary group.
7369 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
7371 * systemctl show-environment gained support for --output=json.
7373 * pam_systemd will now first try to use the X11 abstract socket, and
7374 fallback to the socket file in /tmp/.X11-unix/ only if that does not
7377 * systemd-journald will no longer go back to volatile storage
7378 regardless of configuration when its unit is restarted.
7380 * Initial support for the LoongArch architecture has been added (system
7381 call lists, GPT partition table UUIDs, etc).
7383 * systemd-journald's own logging messages are now also logged to the
7384 journal itself when systemd-journald logs to /dev/kmsg.
7386 * systemd-journald now re-enables COW for archived journal files on
7387 filesystems that support COW. One benefit of this change is that
7388 archived journal files will now get compressed on btrfs filesystems
7389 that have compression enabled.
7391 * systemd-journald now deduplicates fields in a single log message
7392 before adding it to the journal. In archived journal files, it will
7393 also punch holes for unused parts and truncate the file as
7394 appropriate, leading to reductions in disk usage.
7396 * journalctl --verify was extended with more informative error
7399 * More of sd-journal's functions are now resistant against journal file
7402 * The shutdown command learnt a new option --show, to display the
7405 * A LICENSES/ directory is now included in the git tree. It contains a
7406 README.md file that explains the licenses used by source files in
7407 this repository. It also contains the text of all applicable
7408 licenses as they appear on spdx.org.
7410 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
7411 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
7412 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
7413 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
7414 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
7415 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
7416 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
7417 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
7418 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
7419 Christian Brauner, Christian Göttsche, Christian Wehrli,
7420 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
7421 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
7422 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
7423 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
7424 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
7425 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
7426 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
7427 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
7428 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
7429 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
7430 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
7431 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
7432 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
7433 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
7434 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
7435 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
7436 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
7437 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
7438 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
7439 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
7440 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
7441 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
7442 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
7443 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
7444 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
7445 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
7446 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
7447 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
7448 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
7449 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
7450 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
7451 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
7452 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
7453 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
7454 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
7455 Дамјан Георгиевски, наб
7457 — Warsaw, 2021-12-23
7461 * When operating on disk images via the --image= switch of various
7462 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
7463 no 'root=' parameter on the kernel command line, and multiple
7464 suitable root or /usr/ partitions exist in the image, then a simple
7465 comparison inspired by strverscmp() is done on the GPT partition
7466 label, and the newest partition is picked. This permits a simple and
7467 generic whole-file-system A/B update logic where new operating system
7468 versions are dropped into partitions whose label is then updated with
7469 a matching version identifier.
7471 * systemd-sysusers now supports querying the passwords to set for the
7472 users it creates via the "credentials" logic introduced in v247: the
7473 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
7474 credentials are consulted for the password to use (either in UNIX
7475 hashed form, or literally). By default these credentials are inherited
7476 down from PID1 (which in turn imports it from a container manager if
7477 there is one). This permits easy configuration of user passwords
7478 during first boot. Example:
7480 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
7482 Note that systemd-sysusers operates in purely additive mode: it
7483 executes no operation if the declared users already exist, and hence
7484 doesn't set any passwords as effect of the command line above if the
7485 specified root user exists already in the image. (Note that
7486 --volatile=yes ensures it doesn't, though.)
7488 * systemd-firstboot now also supports querying various system
7489 parameters via the credential subsystems. Thus, as above this may be
7490 used to initialize important system parameters on first boot of
7491 previously unprovisioned images (i.e. images with a mostly empty
7494 * PID 1 may now show both the unit name and the unit description
7495 strings in its status output during boot. This may be configured with
7496 StatusUnitFormat=combined in system.conf or
7497 systemd.status-unit-format=combined on the kernel command line.
7499 * The systemd-machine-id-setup tool now supports a --image= switch for
7500 provisioning a machine ID file into an OS disk image, similar to how
7501 --root= operates on an OS file tree. This matches the existing switch
7502 of the same name for systemd-tmpfiles, systemd-firstboot, and
7503 systemd-sysusers tools.
7505 * Similarly, systemd-repart gained support for the --image= switch too.
7506 In combination with the existing --size= option, this makes the tool
7507 particularly useful for easily growing disk images in a single
7508 invocation, following the declarative rules included in the image
7511 * systemd-repart's partition configuration files gained support for a
7512 new switch MakeDirectories= which may be used to create arbitrary
7513 directories inside file systems that are created, before registering
7514 them in the partition table. This is useful in particular for root
7515 partitions to create mount point directories for other partitions
7516 included in the image. For example, a disk image that contains a
7517 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
7518 create /home/ and /var/ as empty directories in the root file system
7519 on its creation, so that the resulting image can be mounted
7520 immediately, even in read-only mode.
7522 * systemd-repart's CopyBlocks= setting gained support for the special
7523 value "auto". If used, a suitable matching partition on the booted OS
7524 is found as source to copy blocks from. This is useful when
7525 implementing replicating installers, that are booted from one medium
7526 and then stream their own root partition onto the target medium.
7528 * systemd-repart's partition configuration files gained support for a
7529 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
7530 GPT partition flags for the created partitions: this is useful for
7531 marking newly created partitions as read-only, or as not being
7532 subject for automatic mounting from creation on.
7534 * The /etc/os-release file has been extended with two new (optional)
7535 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
7536 information for OS images that are updated comprehensively and
7537 atomically as one image. Two new specifiers %M, %A now resolve to
7538 these two fields in the various configuration options that resolve
7541 * portablectl gained a new switch --extension= for enabling portable
7542 service images with extensions that follow the extension image
7543 concept introduced with v248, and thus allows layering multiple
7544 images when setting up the root filesystem of the service.
7546 * systemd-coredump will now extract ELF build-id information from
7547 processes dumping core and include it in the coredump report.
7548 Moreover, it will look for ELF .note.package sections with
7549 distribution packaging meta-information about the crashing process.
7550 This is useful to directly embed the rpm or deb (or any other)
7551 package name and version in ELF files, making it easy to match
7552 coredump reports with the specific package for which the software was
7553 compiled. This is particularly useful on environments with ELF files
7554 from multiple vendors, different distributions and versions, as is
7555 common today in our containerized and sand-boxed world. For further
7558 https://systemd.io/COREDUMP_PACKAGE_METADATA
7560 * A new udev hardware database has been added for FireWire devices
7563 * The "net_id" built-in of udev has been updated with three
7564 backwards-incompatible changes:
7566 - PCI hotplug slot names on s390 systems are now parsed as
7567 hexadecimal numbers. They were incorrectly parsed as decimal
7568 previously, or ignored if the name was not a valid decimal
7571 - PCI onboard indices up to 65535 are allowed. Previously, numbers
7572 above 16383 were rejected. This primarily impacts s390 systems,
7573 where values up to 65535 are used.
7575 - Invalid characters in interface names are replaced with "_".
7577 The new version of the net naming scheme is "v249". The previous
7578 scheme can be selected via the "net.naming_scheme=v247" kernel
7579 command line parameter.
7581 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
7582 NULL bus object, for which they will return false. Or in other words,
7583 an unallocated bus connection is neither ready nor open.
7585 * The sd-device API acquired a new API function
7586 sd_device_get_usec_initialized() that returns the monotonic time when
7587 the udev device first appeared in the database.
7589 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
7590 sd_device_get_trigger_uuid(). The former is similar to
7591 sd_device_trigger() but returns a randomly generated UUID that is
7592 associated with the synthetic uevent generated by the call. This UUID
7593 may be read from the sd_device object a monitor eventually receives,
7594 via the sd_device_get_trigger_uuid(). This interface requires kernel
7595 4.13 or above to work, and allows tracking a synthetic uevent through
7596 the entire device management stack. The "udevadm trigger --settle"
7597 logic has been updated to make use of this concept if available to
7598 wait precisely for the uevents it generates. "udevadm trigger" also
7599 gained a new parameter --uuid that prints the UUID for each generated
7602 * sd-device also gained new APIs sd_device_new_from_ifname() and
7603 sd_device_new_from_ifindex() for allocating an sd-device object for
7604 the specified network interface. The former accepts an interface name
7605 (either a primary or an alternative name), the latter an interface
7608 * The native Journal protocol has been documented. Clients may talk
7609 this as alternative to the classic BSD syslog protocol for locally
7610 delivering log records to the Journal. The protocol has been stable
7611 for a long time and in fact been implemented already in a variety
7612 of alternative client libraries. This documentation makes the support
7615 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
7617 * A new BPFProgram= setting has been added to service files. It may be
7618 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
7619 file, or a bind mount or symlink to one. This may be used to upload
7620 and manage BPF programs externally and then hook arbitrary systemd
7623 * The "home.arpa" domain that has been officially declared as the
7624 choice for domain for local home networks per RFC 8375 has been added
7625 to the default NTA list of resolved, since DNSSEC is generally not
7626 available on private domains.
7628 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
7630 * A new ManageForeignRoutingPolicyRules= setting has been added to
7631 .network files which may be used to exclude foreign-created routing
7632 policy rules from systemd-networkd management.
7634 * systemd-network-wait-online gained two new switches -4 and -6 that
7635 may be used to tweak whether to wait for only IPv4 or only IPv6
7638 * .network files gained a new RequiredFamilyForOnline= setting to
7639 fine-tune whether to require an IPv4 or IPv6 address in order to
7640 consider an interface "online".
7642 * networkctl will now show an over-all "online" state in the per-link
7645 * In .network files a new OutgoingInterface= setting has been added to
7646 specify the output interface in bridge FDB setups.
7648 * In .network files the Multipath group ID may now be configured for
7649 [NextHop] entries, via the new Group= setting.
7651 * The DHCP server logic configured in .network files gained a new
7652 setting RelayTarget= that turns the server into a DHCP server relay.
7653 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
7654 to further tweak the DHCP relay behaviour.
7656 * The DHCP server logic also gained a new ServerAddress= setting in
7657 .network files that explicitly specifies the server IP address to
7658 use. If not specified, the address is determined automatically, as
7661 * The DHCP server logic in systemd-networkd gained support for static
7662 DHCP leases, configurable via the [DHCPServerStaticLease]
7663 section. This allows explicitly mapping specific MAC addresses to
7664 fixed IP addresses and vice versa.
7666 * The RestrictAddressFamilies= setting in service files now supports a
7667 new special value "none". If specified sockets of all address
7668 families will be made unavailable to services configured that way.
7670 * systemd-fstab-generator and systemd-repart have been updated to
7671 support booting from disks that carry only a /usr/ partition but no
7672 root partition yet, and where systemd-repart can add it in on the
7673 first boot. This is useful for implementing systems that ship with a
7674 single /usr/ file system, and whose root file system shall be set up
7675 and formatted on a LUKS-encrypted volume whose key is generated
7676 locally (and possibly enrolled in the TPM) during the first boot.
7678 * The [Address] section of .network files now accepts a new
7679 RouteMetric= setting that configures the routing metric to use for
7680 the prefix route created as effect of the address configuration.
7681 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
7682 gained matching settings for their prefix routes. (The option of the
7683 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
7684 it conceptually belongs there; the old option is still understood for
7687 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
7690 * A new udev property ID_NET_DHCP_BROADCAST on network interface
7691 devices is now honoured by systemd-networkd, controlling whether to
7692 issue DHCP offers via broadcasting. This is used to ensure that s390
7693 layer 3 network interfaces work out-of-the-box with systemd-networkd.
7695 * nss-myhostname and systemd-resolved will now synthesize address
7696 records for a new special hostname "_outbound". The name will always
7697 resolve to the local IP addresses most likely used for outbound
7698 connections towards the default routes. On multi-homed hosts this is
7699 useful to have a stable handle referring to "the" local IP address
7700 that matters most, to the point where this is defined.
7702 * The Discoverable Partition Specification has been updated with a new
7703 GPT partition flag "grow-file-system" defined for its partition
7704 types. Whenever partitions with this flag set are automatically
7705 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
7706 of systemd-nspawn or other tools; and as opposed to explicit mounting
7707 via /etc/fstab), the file system within the partition is
7708 automatically grown to the full size of the partition. If the file
7709 system size already matches the partition size this flag has no
7710 effect. Previously, this functionality has been available via the
7711 explicit x-systemd.growfs mount option, and this new flag extends
7712 this to automatically discovered mounts. A new GrowFileSystem=
7713 setting has been added to systemd-repart drop-in files that allows
7714 configuring this partition flag. This new flag defaults to on for
7715 partitions automatically created by systemd-repart, except if they
7716 are marked read-only. See the specification for further details:
7718 https://systemd.io/DISCOVERABLE_PARTITIONS
7720 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
7721 section. If enabled (which is the default), and an NTP server address
7722 is acquired through a DHCP lease on this interface an explicit route
7723 to this address is created on this interface to ensure that NTP
7724 traffic to the NTP server acquired on an interface is also routed
7725 through that interface. The pre-existing RoutesToDNS= setting that
7726 implements the same for DNS servers is now enabled by default.
7728 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
7729 been added that may be used to restrict the network interfaces
7730 sockets created by the service may be bound to. This is implemented
7733 * A new ConditionFirmware= setting has been added to unit files to
7734 conditionalize on certain firmware features. At the moment it may
7735 check whether running on a UEFI system, a device.tree system, or if
7736 the system is compatible with some specified device-tree feature.
7738 * A new ConditionOSRelease= setting has been added to unit files to
7739 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
7740 operators may be used to check if some field has some specific value
7741 or do an alphanumerical comparison. Equality comparisons are useful
7742 for fields like ID, but relative comparisons for fields like
7743 VERSION_ID or IMAGE_VERSION.
7745 * hostnamed gained a new Describe() D-Bus method that returns a JSON
7746 serialization of the host data it exposes. This is exposed via
7747 "hostnamectl --json=" to acquire a host identity description in JSON.
7748 It's our intention to add a similar features to most services and
7749 objects systemd manages, in order to simplify integration with
7750 program code that can consume JSON.
7752 * Similarly, networkd gained a Describe() method on its Manager and
7753 Link bus objects. This is exposed via "networkctl --json=".
7755 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
7756 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
7757 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
7758 that is used both to get the value (when no argument is given), and
7759 to set the value (when an argument is specified). The old names
7760 continue to be supported for compatibility.
7762 * systemd-detect-virt and ConditionVirtualization= are now able to
7763 correctly identify Amazon EC2 environments.
7765 * The LogLevelMax= setting of unit files now applies not only to log
7766 messages generated *by* the service, but also to log messages
7767 generated *about* the service by PID 1. To suppress logs concerning a
7768 specific service comprehensively, set this option to a high log
7771 * bootctl gained support for a new --make-machine-id-directory= switch
7772 that allows precise control on whether to create the top-level
7773 per-machine directory in the boot partition that typically contains
7774 Type 1 boot loader entries.
7776 * During build SBAT data to include in the systemd-boot EFI PE binaries
7777 may be specified now.
7779 * /etc/crypttab learnt a new option "headless". If specified any
7780 requests to query the user interactively for passwords or PINs will
7781 be skipped. This is useful on systems that are headless, i.e. where
7782 an interactive user is generally not present.
7784 * /etc/crypttab also learnt a new option "password-echo=" that allows
7785 configuring whether the encryption password prompt shall echo the
7786 typed password and if so, do so literally or via asterisks. (The
7787 default is the same behaviour as before: provide echo feedback via
7790 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
7791 systemd-homed has been updated to allow explicit configuration of the
7792 "user presence" and "user verification" checks, as well as whether a
7793 PIN is required for authentication, via the new switches
7794 --fido2-with-user-presence=, --fido2-with-user-verification=,
7795 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
7796 features are available, and may be enabled or disabled depends on the
7799 * systemd-nspawn's --private-user= switch now accepts the special value
7800 "identity" which configures a user namespacing environment with an
7801 identity mapping of 65535 UIDs. This means the container UID 0 is
7802 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
7803 this doesn't appear to be useful, however it does reduce the attack
7804 surface a bit, since the resulting container will possess process
7805 capabilities only within its namespace and not on the host.
7807 * systemd-nspawn's --private-user-chown switch has been replaced by a
7808 more generic --private-user-ownership= switch that accepts one of
7809 three values: "chown" is equivalent to the old --private-user-chown,
7810 and "off" is equivalent to the absence of the old switch. The value
7811 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
7812 of files and directories of the underlying image to the chosen UID
7813 range for the container. "auto" is equivalent to "map" if UID mapping
7814 mount are supported, otherwise it is equivalent to "chown". The short
7815 -U switch systemd-nspawn now implies --private-user-ownership=auto
7816 instead of the old --private-user-chown. Effectively this means: if
7817 the backing file system supports UID mapping mounts the feature is
7818 now used by default if -U is used. Generally, it's a good idea to use
7819 UID mapping mounts instead of recursive chown()ing, since it allows
7820 running containers off immutable images (since no modifications of
7821 the images need to take place), and share images between multiple
7822 instances. Moreover, the recursive chown()ing operation is slow and
7823 can be avoided. Conceptually it's also a good thing if transient UID
7824 range uses do not leak into persistent file ownership anymore. TLDR:
7825 finally, the last major drawback of user namespacing has been
7826 removed, and -U should always be used (unless you use btrfs, where
7827 UID mapped mounts do not exist; or your container actually needs
7828 privileges on the host).
7830 * nss-systemd now synthesizes user and group shadow records in addition
7831 to the main user and group records. Thus, hashed passwords managed by
7832 systemd-homed are now accessible via the shadow database.
7834 * The userdb logic (and thus nss-systemd, and so on) now read
7835 additional user/group definitions in JSON format from the drop-in
7836 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
7837 /usr/lib/userdb/. This is a simple and powerful mechanism for making
7838 additional users available to the system, with full integration into
7839 NSS including the shadow databases. Since the full JSON user/group
7840 record format is supported this may also be used to define users with
7841 resource management settings and other runtime settings that
7842 pam_systemd and systemd-logind enforce at login.
7844 * The userdbctl tool gained two new switches --with-dropin= and
7845 --with-varlink= which can be used to fine-tune the sources used for
7846 user database lookups.
7848 * systemd-nspawn gained a new switch --bind-user= for binding a host
7849 user account into the container. This does three things: the user's
7850 home directory is bind mounted from the host into the container,
7851 below the /run/userdb/home/ hierarchy. A free UID is picked in the
7852 container, and a user namespacing UID mapping to the host user's UID
7853 installed. And finally, a minimal JSON user and group record (along
7854 with its hashed password) is dropped into /run/host/userdb/. These
7855 records are picked up automatically by the userdb drop-in logic
7856 describe above, and allow the user to login with the same password as
7857 on the host. Effectively this means: if host and container run new
7858 enough systemd versions making a host user available to the container
7859 is trivially simple.
7861 * systemd-journal-gatewayd now supports the switches --user, --system,
7862 --merge, --file= that are equivalent to the same switches of
7863 journalctl, and permit exposing only the specified subset of the
7866 * The OnFailure= dependency between units is now augmented with a
7867 implicit reverse dependency OnFailureOf= (this new dependency cannot
7868 be configured directly it's only created as effect of an OnFailure=
7869 dependency in the reverse order — it's visible in "systemctl show"
7870 however). Similar, Slice= now has an reverse dependency SliceOf=,
7871 that is also not configurable directly, but useful to determine all
7872 units that are members of a slice.
7874 * A pair of new dependency types between units PropagatesStopTo= +
7875 StopPropagatedFrom= has been added, that allows propagation of unit
7876 stop events between two units. It operates similar to the existing
7877 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
7879 * A new dependency type OnSuccess= has been added (plus the reverse
7880 dependency OnSuccessOf=, which cannot be configured directly, but
7881 exists only as effect of the reverse OnSuccess=). It is similar to
7882 OnFailure=, but triggers in the opposite case: when a service exits
7883 cleanly. This allows "chaining up" of services where one or more
7884 services are started once another service has successfully completed.
7886 * A new dependency type Upholds= has been added (plus the reverse
7887 dependency UpheldBy=, which cannot be configured directly, but exists
7888 only as effect of Upholds=). This dependency type is a stronger form
7889 of Wants=: if a unit has an UpHolds= dependency on some other unit
7890 and the former is active then the latter is started whenever it is
7891 found inactive (and no job is queued for it). This is an alternative
7892 to Restart= inside service units, but less configurable, and the
7893 request to uphold a unit is not encoded in the unit itself but in
7894 another unit that intends to uphold it.
7896 * The systemd-ask-password tool now also supports reading passwords
7897 from the credentials subsystem, via the new --credential= switch.
7899 * The systemd-ask-password tool learnt a new switch --emoji= which may
7900 be used to explicit control whether the lock and key emoji (🔐) is
7901 shown in the password prompt on suitable TTYs.
7903 * The --echo switch of systemd-ask-password now optionally takes a
7904 parameter that controls character echo. It may either show asterisks
7905 (default, as before), turn echo off entirely, or echo the typed
7906 characters literally.
7908 * The systemd-ask-password tool also gained a new -n switch for
7909 suppressing output of a trailing newline character when writing the
7910 acquired password to standard output, similar to /bin/echo's -n
7913 * New documentation has been added that describes the organization of
7914 the systemd source code tree:
7916 https://systemd.io/ARCHITECTURE
7918 * Units using ConditionNeedsUpdate= will no longer be activated in
7921 * It is now possible to list a template unit in the WantedBy= or
7922 RequiredBy= settings of the [Install] section of another template
7923 unit, which will be instantiated using the same instance name.
7925 * A new MemoryAvailable property is available for units. If the unit,
7926 or the slices it is part of, have a memory limit set via MemoryMax=/
7927 MemoryHigh=, MemoryAvailable will indicate how much more memory the
7928 unit can claim before hitting the limits.
7930 * systemd-coredump will now try to stay below the cgroup memory limit
7931 placed on itself or one of the slices it runs under, if the storage
7932 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
7933 since files written on such filesystems count toward the cgroup memory
7934 limit. If there is not enough available memory in such cases to store
7935 the core file uncompressed, systemd-coredump will skip to compressed
7936 storage directly (if enabled) and it will avoid analyzing the core file
7937 to print backtrace and metadata in the journal.
7939 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
7940 of a path matches the configured expectations, and remove it if not.
7942 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
7943 specify which of the several available filesystem timestamps (access
7944 time, birth time, change time, modification time) to look at when
7945 deciding whether a path has aged enough to be cleaned.
7947 * A new IPv6StableSecretAddress= setting has been added to .network
7948 files, which takes an IPv6 address to use as secret for IPv6 address
7951 * The [DHCPServer] logic in .network files gained support for a new
7952 UplinkInterface= setting that permits configuration of the uplink
7953 interface name to propagate DHCP lease information from.
7955 * The WakeOnLan= setting in .link files now accepts a list of flags
7956 instead of a single one, to configure multiple wake-on-LAN policies.
7958 * User-space defined tracepoints (USDT) have been added to udev at
7959 strategic locations. This is useful for tracing udev behaviour and
7960 performance with bpftrace and similar tools.
7962 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
7963 setting a network timeout time.
7965 * If a system service is running in a new mount namespace (RootDirectory=
7966 and friends), all file systems will be mounted with MS_NOSUID by
7967 default, unless the system is running with SELinux enabled.
7969 * When enumerating time zones the timedatectl tool will now consult the
7970 'tzdata.zi' file shipped by the IANA time zone database package, in
7971 addition to 'zone1970.tab', as before. This makes sure time zone
7972 aliases are now correctly supported. Some distributions so far did
7973 not install this additional file, most do however. If you
7974 distribution does not install it yet, it might make sense to change
7977 * Intel HID rfkill event is no longer masked, since it's the only
7978 source of rfkill event on newer HP laptops. To have both backward and
7979 forward compatibility, userspace daemon needs to debounce duplicated
7980 events in a short time window.
7982 Contributions from: Aakash Singh, adrian5, Albert Brox,
7983 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
7984 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
7985 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
7986 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
7987 Christian Hesse, Daniel Schaefer, Dan Streetman,
7988 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
7989 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
7990 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
7991 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
7992 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
7993 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
7994 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
7995 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
7996 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
7997 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
7998 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
7999 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
8000 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
8001 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
8002 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
8003 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
8004 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
8005 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
8006 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
8007 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
8008 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
8009 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
8010 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
8011 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
8012 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
8014 — Edinburgh, 2021-07-07
8018 * A concept of system extension images is introduced. Such images may
8019 be used to extend the /usr/ and /opt/ directory hierarchies at
8020 runtime with additional files (even if the file system is read-only).
8021 When a system extension image is activated, its /usr/ and /opt/
8022 hierarchies and os-release information are combined via overlayfs
8023 with the file system hierarchy of the host OS.
8025 A new systemd-sysext tool can be used to merge, unmerge, list, and
8026 refresh system extension hierarchies. See
8027 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
8029 The systemd-sysext.service automatically merges installed system
8030 extensions during boot (before basic.target, but not in very early
8031 boot, since various file systems have to be mounted first).
8033 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
8034 supported system extension level.
8036 * A new ExtensionImages= unit setting can be used to apply the same
8037 system extension image concept from systemd-sysext to the namespaced
8038 file hierarchy of specific services, following the same rules and
8041 * Support for a new special "root=tmpfs" kernel command-line option has
8042 been added. When specified, a tmpfs is mounted on /, and mount.usr=
8043 should be used to point to the operating system implementation.
8045 * A new configuration file /etc/veritytab may be used to configure
8046 dm-verity integrity protection for block devices. Each line is in the
8047 format "volume-name data-device hash-device roothash options",
8048 similar to /etc/crypttab.
8050 * A new kernel command-line option systemd.verity.root_options= may be
8051 used to configure dm-verity behaviour for the root device.
8053 * The key file specified in /etc/crypttab (the third field) may now
8054 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
8055 acquired by connecting to that socket and reading from it. This
8056 allows the implementation of a service to provide key information
8057 dynamically, at the moment when it is needed.
8059 * When the hostname is set explicitly to "localhost", systemd-hostnamed
8060 will respect this. Previously such a setting would be mostly silently
8061 ignored. The goal is to honour configuration as specified by the
8064 * The fallback hostname that will be used by the system manager and
8065 systemd-hostnamed can now be configured in two new ways: by setting
8066 DEFAULT_HOSTNAME= in os-release(5), or by setting
8067 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
8068 also be configured during compilation. The environment variable is
8069 intended for testing and local overrides, the os-release(5) field is
8070 intended to allow customization by different variants of a
8071 distribution that share the same compiled packages.
8073 * The environment block of the manager itself may be configured through
8074 a new ManagerEnvironment= setting in system.conf or user.conf. This
8075 complements existing ways to set the environment block (the kernel
8076 command line for the system manager, the inherited environment and
8077 user@.service unit file settings for the user manager).
8079 * systemd-hostnamed now exports the default hostname and the source of
8080 the configured hostname ("static", "transient", or "default") as
8083 * systemd-hostnamed now exports the "HardwareVendor" and
8084 "HardwareModel" D-Bus properties, which are supposed to contain a
8085 pair of cleaned up, human-readable strings describing the system's
8086 vendor and model. It's typically sourced from the firmware's DMI
8087 tables, but may be augmented from a new hwdb database. hostnamectl
8088 shows this in the status output.
8090 * Support has been added to systemd-cryptsetup for extracting the
8091 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
8092 metadata header. This allows the information how to open the
8093 encrypted device to be embedded directly in the device and obviates
8094 the need for configuration in an external file.
8096 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
8097 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
8098 pre-existing support for PKCS#11 security tokens).
8100 * systemd-repart may enroll encrypted partitions using TPM2
8101 hardware. This may be useful for example to create an encrypted /var
8102 partition bound to the machine on first boot.
8104 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
8105 and PKCS#11 security tokens to LUKS volumes, list and destroy
8108 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
8110 It also supports enrolling "recovery keys" and regular passphrases.
8112 * The libfido2 dependency is now based on dlopen(), so that the library
8113 is used at runtime when installed, but is not a hard runtime
8116 * systemd-cryptsetup gained support for two new options in
8117 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
8118 request synchronous processing of encryption/decryption IO.
8120 * The manager may be configured at compile time to use the fexecve()
8121 instead of the execve() system call when spawning processes. Using
8122 fexecve() closes a window between checking the security context of an
8123 executable and spawning it, but unfortunately the kernel displays
8124 stale information in the process' "comm" field, which impacts ps
8127 * The configuration option -Dcompat-gateway-hostname has been dropped.
8128 "_gateway" is now the only supported name.
8130 * The ConditionSecurity=tpm2 unit file setting may be used to check if
8131 the system has at least one TPM2 (tpmrm class) device.
8133 * A new ConditionCPUFeature= has been added that may be used to
8134 conditionalize units based on CPU features. For example,
8135 ConditionCPUFeature=rdrand will condition a unit so that it is only
8136 run when the system CPU supports the RDRAND opcode.
8138 * The existing ConditionControlGroupController= setting has been
8139 extended with two new values "v1" and "v2". "v2" means that the
8140 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
8141 hierarchy or the hybrid hierarchy are used.
8143 * A new PrivateIPC= setting on a unit file allows executed processes to
8144 be moved into a private IPC namespace, with separate System V IPC
8145 identifiers and POSIX message queues.
8147 A new IPCNamespacePath= allows the unit to be joined to an existing
8150 * The tables of system calls in seccomp filters are now automatically
8151 generated from kernel lists exported on
8152 https://fedora.juszkiewicz.com.pl/syscalls.html.
8154 The following architectures should now have complete lists:
8155 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
8156 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
8158 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
8159 on /run/ if it is not already a mount point. A writable /run/ has
8160 always been a requirement for a functioning system, but this was not
8161 guaranteed when using a read-only image.
8163 Users can always specify BindPaths= or InaccessiblePaths= as
8164 overrides, and they will take precedence. If the host's root mount
8165 point is used, there is no change in behaviour.
8167 * New bind mounts and file system image mounts may be injected into the
8168 mount namespace of a service (without restarting it). This is exposed
8169 respectively as 'systemctl bind <unit> <path>…' and
8170 'systemctl mount-image <unit> <image>…'.
8172 * The StandardOutput= and StandardError= settings can now specify files
8173 to be truncated for output (as "truncate:<path>").
8175 * The ExecPaths= and NoExecPaths= settings may be used to specify
8176 noexec for parts of the file system.
8178 * sd-bus has a new function sd_bus_open_user_machine() to open a
8179 connection to the session bus of a specific user in a local container
8180 or on the local host. This is exposed in the existing -M switch to
8181 systemctl and similar tools:
8183 systemctl --user -M lennart@foobar start foo
8185 This will connect to the user bus of a user "lennart" in container
8186 "foobar". If no container name is specified, the specified user on
8187 the host itself is connected to
8189 systemctl --user -M lennart@ start quux
8191 * sd-bus also gained a convenience function sd_bus_message_send() to
8192 simplify invocations of sd_bus_send(), taking only a single
8193 parameter: the message to send.
8195 * sd-event allows rate limits to be set on event sources, for dealing
8196 with high-priority event sources that might starve out others. See
8197 the new man page sd_event_source_set_ratelimit(3) for details.
8199 * systemd.link files gained a [Link] Promiscuous= switch, which allows
8200 the device to be raised in promiscuous mode.
8202 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
8203 number of TX and RX queues to be configured.
8205 New [Link] TransmitQueueLength= setting allows the size of the TX
8206 queue to be configured.
8208 New [Link] GenericSegmentOffloadMaxBytes= and
8209 GenericSegmentOffloadMaxSegments= allow capping the packet size and
8210 the number of segments accepted in Generic Segment Offload.
8212 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
8213 wireless routing protocol that operates on ISO/OSI Layer 2 only and
8214 uses ethernet frames to route/bridge packets. This encompasses a new
8215 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
8216 new settings in .netdev files, and a new BatmanAdvanced= setting in
8219 * systemd.network files gained a [Network] RouteTable= configuration
8220 switch to select the routing policy table.
8222 systemd.network files gained a [RoutingPolicyRule] Type=
8223 configuration switch (one of "blackhole, "unreachable", "prohibit").
8225 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
8226 RouteAllowList= settings to ignore/accept route advertisements from
8227 routers matching specified prefixes. The DenyList= setting has been
8228 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
8231 systemd.network files gained a [DHCPv6] UseAddress= setting to
8232 optionally ignore the address provided in the lease.
8234 systemd.network files gained a [DHCPv6PrefixDelegation]
8235 ManageTemporaryAddress= switch.
8237 systemd.network files gained a new ActivationPolicy= setting which
8238 allows configuring how the UP state of an interface shall be managed,
8239 i.e. whether the interface is always upped, always downed, or may be
8240 upped/downed by the user using "ip link set dev".
8242 * The default for the Broadcast= setting in .network files has slightly
8243 changed: the broadcast address will not be configured for wireguard
8246 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
8247 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
8248 configuration options for VLAN packet handling.
8250 * udev rules may now set log_level= option. This allows debug logs to
8251 be enabled for select events, e.g. just for a specific subsystem or
8252 even a single device.
8254 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
8255 DATA_PREPARED_ID properties for block devices with ISO9660 file
8258 * udev now exports decoded DMI information about installed memory slots
8259 as device properties under the /sys/class/dmi/id/ pseudo device.
8261 * /dev/ is not mounted noexec anymore. This didn't provide any
8262 significant security benefits and would conflict with the executable
8263 mappings used with /dev/sgx device nodes. The previous behaviour can
8264 be restored for individual services with NoExecPaths=/dev (or by allow-
8265 listing and excluding /dev from ExecPaths=).
8267 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
8268 and /dev/vhost-net are owned by the kvm group.
8270 * The hardware database has been extended with a list of fingerprint
8271 readers that correctly support USB auto-suspend using data from
8274 * systemd-resolved can now answer DNSSEC questions through the stub
8275 resolver interface in a way that allows local clients to do DNSSEC
8276 validation themselves. For a question with DO+CD set, it'll proxy the
8277 DNS query and respond with a mostly unmodified packet received from
8278 the upstream server.
8280 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
8281 resolved.conf. If true the service will provide caching even for DNS
8282 lookups made to an upstream DNS server on the 127.0.0.1/::1
8283 addresses. By default (and when the option is false) systemd-resolved
8284 will not cache such lookups, in order to avoid duplicate local
8285 caching, under the assumption the local upstream server caches
8288 * systemd-resolved now implements RFC5001 NSID in its local DNS
8289 stub. This may be used by local clients to determine whether they are
8290 talking to the DNS resolver stub or a different DNS server.
8292 * When resolving host names and other records resolvectl will now
8293 report where the data was acquired from (i.e. the local cache, the
8294 network, locally synthesized, …) and whether the network traffic it
8295 affected was encrypted or not. Moreover the tool acquired a number of
8296 new options --cache=, --synthesize=, --network=, --zone=,
8297 --trust-anchor=, --validate= that take booleans and may be used to
8298 tweak a lookup, i.e. whether it may be answered from cached
8299 information, locally synthesized information, information acquired
8300 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
8301 anchor, and whether DNSSEC validation shall be executed for the
8304 * systemd-nspawn gained a new --ambient-capability= setting
8305 (AmbientCapability= in .nspawn files) to configure ambient
8306 capabilities passed to the container payload.
8308 * systemd-nspawn gained the ability to configure the firewall using the
8309 nftables subsystem (in addition to the existing iptables support).
8310 Similarly, systemd-networkd's IPMasquerade= option now supports
8311 nftables as back-end, too. In both cases NAT on IPv6 is now supported
8312 too, in addition to IPv4 (the iptables back-end still is IPv4-only).
8314 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
8315 retains its meaning, but has been deprecated. Please switch to either
8316 "ivp4" or "both" (if covering IPv6 is desired).
8318 * systemd-importd will now download .verity and .roothash.p7s files
8319 along with the machine image (as exposed via machinectl pull-raw).
8321 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
8322 setting to configure the time a unit's cgroup needs to exceed memory
8323 pressure limits before action will be taken, and a new
8324 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
8327 systemd-oomd is now considered fully supported (the usual
8328 backwards-compatibility promises apply). Swap is not required for
8329 operation, but it is still recommended.
8331 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
8332 configures the retry delay when trying to contact servers.
8334 * systemd-stdio-bridge gained --system/--user options to connect to the
8335 system bus (previous default) or the user session bus.
8337 * systemd-localed may now call locale-gen to generate missing locales
8338 on-demand (UTF-8-only). This improves integration with Debian-based
8339 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
8341 * systemctl --check-inhibitors=true may now be used to obey inhibitors
8342 even when invoked non-interactively. The old --ignore-inhibitors
8343 switch is now deprecated and replaced by --check-inhibitors=false.
8345 * systemctl import-environment will now emit a warning when called
8346 without any arguments (i.e. to import the full environment block of
8347 the called program). This command will usually be invoked from a
8348 shell, which means that it'll inherit a bunch of variables which are
8349 specific to that shell, and usually to the TTY the shell is connected
8350 to, and don't have any meaning in the global context of the system or
8351 user service manager. Instead, only specific variables should be
8352 imported into the manager environment block.
8354 Similarly, programs which update the manager environment block by
8355 directly calling the D-Bus API of the manager, should also push
8356 specific variables, and not the full inherited environment.
8358 * systemctl's status output now shows unit state with a more careful
8359 choice of Unicode characters: units in maintenance show a "○" symbol
8360 instead of the usual "●", failed units show "×", and services being
8363 * coredumpctl gained a --debugger-arguments= switch to pass arguments
8364 to the debugger. It also gained support for showing coredump info in
8365 a simple JSON format.
8367 * systemctl/loginctl/machinectl's --signal= option now accept a special
8368 value "list", which may be used to show a brief table with known
8369 process signals and their numbers.
8371 * networkctl now shows the link activation policy in status.
8373 * Various tools gained --pager/--no-pager/--json= switches to
8374 enable/disable the pager and provide JSON output.
8376 * Various tools now accept two new values for the SYSTEMD_COLORS
8377 environment variable: "16" and "256", to configure how many terminal
8378 colors are used in output.
8380 * less 568 or newer is now required for the auto-paging logic of the
8381 various tools. Hyperlink ANSI sequences in terminal output are now
8382 used even if a pager is used, and older versions of less are not able
8383 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
8384 disable this output again.
8386 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
8387 builds, non-merged-usr builds) are now officially deprecated. A
8388 warning is emitted during build. Support is slated to be removed in
8389 about a year (when the Debian Bookworm release development starts).
8391 * Systems with the legacy cgroup v1 hierarchy are now marked as
8392 "tainted", to make it clearer that using the legacy hierarchy is not
8395 * systemd-localed will now refuse to configure a keymap which is not
8396 installed in the file system. This is intended as a bug fix, but
8397 could break cases where systemd-localed was used to configure the
8398 keymap in advanced of it being installed. It is necessary to install
8399 the keymap file first.
8401 * The main git development branch has been renamed to 'main'.
8403 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
8404 for partitions, as in the vast majority of cases they contain none
8405 and are used internally by the bootloader (eg: uboot).
8407 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
8408 spawned processes to the PID of the process itself. This may be used
8409 by programs for detecting whether they were forked off by the service
8410 manager itself or are a process forked off further down the tree.
8412 * The sd-device API gained four new calls: sd_device_get_action() to
8413 determine the uevent add/remove/change/… action the device object has
8414 been seen for, sd_device_get_seqno() to determine the uevent sequence
8415 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
8416 object from stat(2) data of a device node, and sd_device_trigger() to
8417 write to the 'uevent' attribute of a device.
8419 * For most tools the --no-legend= switch has been replaced by
8420 --legend=no and --legend=yes, to force whether tables are shown with
8423 * Units acquired a new property "Markers" that takes a list of zero,
8424 one or two of the following strings: "needs-reload" and
8425 "needs-restart". These markers may be set via "systemctl
8426 set-property". Once a marker is set, "systemctl reload-or-restart
8427 --marked" may be invoked to execute the operation the units are
8428 marked for. This is useful for package managers that want to mark
8429 units for restart/reload while updating, but affect the actual
8430 operations at a later step at once.
8432 * The sd_bus_message_read_strv() API call of sd-bus may now also be
8433 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
8436 * bootctl will now report whether the UEFI firmware used a TPM2 device
8437 and measured the boot process into it.
8439 * systemd-tmpfiles learnt support for a new environment variable
8440 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
8441 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
8442 even if the root fs of the system is not itself a btrfs volume.
8444 * systemd-detect-virt/ConditionVirtualization= will now explicitly
8445 detect Docker/Podman environments where possible. Moreover, they
8446 should be able to generically detect any container manager as long as
8447 it assigns the container a cgroup.
8449 * portablectl gained a new "reattach" verb for detaching/reattaching a
8450 portable service image, useful for updating images on-the-fly.
8452 * Intel SGX enclave device nodes (which expose a security feature of
8453 newer Intel CPUs) will now be owned by a new system group "sgx".
8455 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
8456 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
8457 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
8458 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
8459 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
8460 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
8461 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
8462 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
8463 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
8464 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
8465 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
8466 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
8467 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
8468 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
8469 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
8470 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
8471 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
8472 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
8473 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
8474 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
8475 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
8476 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
8477 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
8478 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
8479 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
8480 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
8481 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
8482 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
8483 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
8484 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
8485 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
8486 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
8487 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
8488 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
8489 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
8490 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
8491 Zmicer Turok, Дамјан Георгиевски
8493 — Berlin, 2021-03-30
8497 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
8498 "bind" and "unbind" to the Linux device model. When this kernel
8499 change was made, systemd-udevd was only minimally updated to handle
8500 and propagate these new event types. The introduction of these new
8501 uevents (which are typically generated for USB devices and devices
8502 needing a firmware upload before being functional) resulted in a
8503 number of issues which we so far didn't address. We hoped the kernel
8504 maintainers would themselves address these issues in some form, but
8505 that did not happen. To handle them properly, many (if not most) udev
8506 rules files shipped in various packages need updating, and so do many
8507 programs that monitor or enumerate devices with libudev or sd-device,
8508 or otherwise process uevents. Please note that this incompatibility
8509 is not fault of systemd or udev, but caused by an incompatible kernel
8510 change that happened back in Linux 4.14, but is becoming more and
8511 more visible as the new uevents are generated by more kernel drivers.
8513 To minimize issues resulting from this kernel change (but not avoid
8514 them entirely) starting with systemd-udevd 247 the udev "tags"
8515 concept (which is a concept for marking and filtering devices during
8516 enumeration and monitoring) has been reworked: udev tags are now
8517 "sticky", meaning that once a tag is assigned to a device it will not
8518 be removed from the device again until the device itself is removed
8519 (i.e. unplugged). This makes sure that any application monitoring
8520 devices that match a specific tag is guaranteed to both see uevents
8521 where the device starts being relevant, and those where it stops
8522 being relevant (the latter now regularly happening due to the new
8523 "unbind" uevent type). The udev tags concept is hence now a concept
8524 tied to a *device* instead of a device *event* — unlike for example
8525 udev properties whose lifecycle (as before) is generally tied to a
8526 device event, meaning that the previously determined properties are
8527 forgotten whenever a new uevent is processed.
8529 With the newly redefined udev tags concept, sometimes it's necessary
8530 to determine which tags are the ones applied by the most recent
8531 uevent/database update, in order to discern them from those
8532 originating from earlier uevents/database updates of the same
8533 device. To accommodate for this a new automatic property CURRENT_TAGS
8534 has been added that works similar to the existing TAGS property but
8535 only lists tags set by the most recent uevent/database
8536 update. Similarly, the libudev/sd-device API has been updated with
8537 new functions to enumerate these 'current' tags, in addition to the
8538 existing APIs that now enumerate the 'sticky' ones.
8540 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
8541 essential that all udev rules files and applications are updated to
8542 handle the new events. Specifically:
8544 • All rule files that currently use a header guard similar to
8545 ACTION!="add|change",GOTO="xyz_end" should be updated to use
8546 ACTION=="remove",GOTO="xyz_end" instead, so that the
8547 properties/tags they add are also applied whenever "bind" (or
8548 "unbind") is seen. (This is most important for all physical device
8549 types — those for which "bind" and "unbind" are currently
8550 generated, for all other device types this change is still
8551 recommended but not as important — but certainly prepares for
8552 future kernel uevent type additions).
8554 • Similarly, all code monitoring devices that contains an 'if' branch
8555 discerning the "add" + "change" uevent actions from all other
8556 uevents actions (i.e. considering devices only relevant after "add"
8557 or "change", and irrelevant on all other events) should be reworked
8558 to instead negatively check for "remove" only (i.e. considering
8559 devices relevant after all event types, except for "remove", which
8560 invalidates the device). Note that this also means that devices
8561 should be considered relevant on "unbind", even though conceptually
8562 this — in some form — invalidates the device. Since the precise
8563 effect of "unbind" is not generically defined, devices should be
8564 considered relevant even after "unbind", however I/O errors
8565 accessing the device should then be handled gracefully.
8567 • Any code that uses device tags for deciding whether a device is
8568 relevant or not most likely needs to be updated to use the new
8569 udev_device_has_current_tag() API (or sd_device_has_current_tag()
8570 in case sd-device is used), to check whether the tag is set at the
8571 moment an uevent is seen (as opposed to the existing
8572 udev_device_has_tag() API which checks if the tag ever existed on
8573 the device, following the API concept redefinition explained
8576 We are very sorry for this breakage and the requirement to update
8577 packages using these interfaces. We'd again like to underline that
8578 this is not caused by systemd/udev changes, but result of a kernel
8581 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
8582 packages have not retriggered devices once the udev package (or any
8583 auxiliary package installing additional udev rules) is updated. We
8584 intend to work with major distributions to change this, so that
8585 "udevadm trigger -c change" is issued on such upgrades, ensuring that
8586 the updated ruleset is applied to the devices already discovered, so
8587 that (asynchronously) after the upgrade completed the udev database
8588 is consistent with the updated rule set. This means udev rules must
8589 be ready to be retriggered with a "change" action any time, and
8590 result in correct and complete udev database entries. While the
8591 majority of udev rule files known to us currently get this right,
8592 some don't. Specifically, there are udev rules files included in
8593 various packages that only set udev properties on the "add" action,
8594 but do not handle the "change" action. If a device matching those
8595 rules is retriggered with the "change" action (as is intended here)
8596 it would suddenly lose the relevant properties. This always has been
8597 problematic, but as soon as all udev devices are triggered on relevant
8598 package upgrades this will become particularly so. It is strongly
8599 recommended to fix offending rules so that they can handle a "change"
8600 action at any time, and acquire all necessary udev properties even
8601 then. Or in other words: the header guard mentioned above
8602 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
8603 this, as it makes sure rules are rerun on "change" correctly, and
8604 accumulate the correct and complete set of udev properties. udev rule
8605 definitions that cannot handle "change" events being triggered at
8606 arbitrary times should be considered buggy.
8608 * The MountAPIVFS= service file setting now defaults to on if
8609 RootImage= and RootDirectory= are used, which means that with those
8610 two settings /proc/, /sys/ and /dev/ are automatically properly set
8611 up for services. Previous behaviour may be restored by explicitly
8612 setting MountAPIVFS=off.
8614 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
8615 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
8616 latter it takes precedence over the former, similar to how most of
8617 systemd's own configuration is handled. Given that PAM stack
8618 definitions are primarily put together by OS vendors/distributions
8619 (though possibly overridden by users), this systemd release moves its
8620 own PAM stack configuration for the "systemd-user" PAM service (i.e.
8621 for the PAM session invoked by the per-user user@.service instance)
8622 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
8623 packages' vendor versions of their PAM stack definitions from
8624 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
8625 desired the location to which systemd installs its PAM stack
8626 configuration may be changed via the -Dpamconfdir Meson option.
8628 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
8629 libpwquality and libcryptsetup have been changed to be based on
8630 dlopen(): instead of regular dynamic library dependencies declared in
8631 the binary ELF headers, these libraries are now loaded on demand
8632 only, if they are available. If the libraries cannot be found the
8633 relevant operations will fail gracefully, or a suitable fallback
8634 logic is chosen. This is supposed to be useful for general purpose
8635 distributions, as it allows minimizing the list of dependencies the
8636 systemd packages pull in, permitting building of more minimal OS
8637 images, while still making use of these "weak" dependencies should
8638 they be installed. Since many package managers automatically
8639 synthesize package dependencies from ELF shared library dependencies,
8640 some additional manual packaging work has to be done now to replace
8641 those (slightly downgraded from "required" to "recommended" or
8642 whatever is conceptually suitable for the package manager). Note that
8643 this change does not alter build-time behaviour: as before the
8644 build-time dependencies have to be installed during build, even if
8645 they now are optional during runtime.
8647 * sd-event.h gained a new call sd_event_add_time_relative() for
8648 installing timers relative to the current time. This is mostly a
8649 convenience wrapper around the pre-existing sd_event_add_time() call
8650 which installs absolute timers.
8652 * sd-event event sources may now be placed in a new "exit-on-failure"
8653 mode, which may be controlled via the new
8654 sd_event_source_get_exit_on_failure() and
8655 sd_event_source_set_exit_on_failure() functions. If enabled, any
8656 failure returned by the event source handler functions will result in
8657 exiting the event loop (unlike the default behaviour of just
8658 disabling the event source but continuing with the event loop). This
8659 feature is useful to set for all event sources that define "primary"
8660 program behaviour (where failure should be fatal) in contrast to
8661 "auxiliary" behaviour (where failure should remain local).
8663 * Most event source types sd-event supports now accept a NULL handler
8664 function, in which case the event loop is exited once the event
8665 source is to be dispatched, using the userdata pointer — converted to
8666 a signed integer — as exit code of the event loop. Previously this
8667 was supported for IO and signal event sources already. Exit event
8668 sources still do not support this (simply because it makes little
8669 sense there, as the event loop is already exiting when they are
8672 * A new per-unit setting RootImageOptions= has been added which allows
8673 tweaking the mount options for any file system mounted as effect of
8674 the RootImage= setting.
8676 * Another new per-unit setting MountImages= has been added, that allows
8677 mounting additional disk images into the file system tree accessible
8680 * Timer units gained a new FixedRandomDelay= boolean setting. If
8681 enabled, the random delay configured with RandomizedDelaySec= is
8682 selected in a way that is stable on a given system (though still
8683 different for different units).
8685 * Socket units gained a new setting Timestamping= that takes "us", "ns"
8686 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
8689 * systemd-repart now generates JSON output when requested with the new
8692 * systemd-machined's OpenMachineShell() bus call will now pass
8693 additional policy metadata data fields to the PolicyKit
8694 authentication request.
8696 * systemd-tmpfiles gained a new -E switch, which is equivalent to
8697 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
8698 --exclude=/sys. It's particularly useful in combination with --root=,
8699 when operating on OS trees that do not have any of these four runtime
8700 directories mounted, as this means no files below these subtrees are
8701 created or modified, since those mount points should probably remain
8704 * systemd-tmpfiles gained a new --image= switch which is like --root=,
8705 but takes a disk image instead of a directory as argument. The
8706 specified disk image is mounted inside a temporary mount namespace
8707 and the tmpfiles.d/ drop-ins stored in the image are executed and
8708 applied to the image. systemd-sysusers similarly gained a new
8709 --image= switch, that allows the sysusers.d/ drop-ins stored in the
8710 image to be applied onto the image.
8712 * Similarly, the journalctl command also gained an --image= switch,
8713 which is a quick one-step solution to look at the log data included
8716 * journalctl's --output=cat option (which outputs the log content
8717 without any metadata, just the pure text messages) will now make use
8718 of terminal colors when run on a suitable terminal, similarly to the
8721 * JSON group records now support a "description" string that may be
8722 used to add a human-readable textual description to such groups. This
8723 is supposed to match the user's GECOS field which traditionally
8724 didn't have a counterpart for group records.
8726 * The "systemd-dissect" tool that may be used to inspect OS disk images
8727 and that was previously installed to /usr/lib/systemd/ has now been
8728 moved to /usr/bin/, reflecting its updated status of an officially
8729 supported tool with a stable interface. It gained support for a new
8730 --mkdir switch which when combined with --mount has the effect of
8731 creating the directory to mount the image to if it is missing
8732 first. It also gained two new commands --copy-from and --copy-to for
8733 copying files and directories in and out of an OS image without the
8734 need to manually mount it. It also acquired support for a new option
8735 --json= to generate JSON output when inspecting an OS image.
8737 * The cgroup2 file system is now mounted with the
8738 "memory_recursiveprot" mount option, supported since kernel 5.7. This
8739 means that the MemoryLow= and MemoryMin= unit file settings now apply
8740 recursively to whole subtrees.
8742 * systemd-homed now defaults to using the btrfs file system — if
8743 available — when creating home directories in LUKS volumes. This may
8744 be changed with the DefaultFileSystemType= setting in homed.conf.
8745 It's now the default file system in various major distributions and
8746 has the major benefit for homed that it can be grown and shrunk while
8747 mounted, unlike the other contenders ext4 and xfs, which can both be
8748 grown online, but not shrunk (in fact xfs is the technically most
8749 limited option here, as it cannot be shrunk at all).
8751 * JSON user records managed by systemd-homed gained support for
8752 "recovery keys". These are basically secondary passphrases that can
8753 unlock user accounts/home directories. They are computer-generated
8754 rather than user-chosen, and typically have greater entropy.
8755 homectl's --recovery-key= option may be used to add a recovery key to
8756 a user account. The generated recovery key is displayed as a QR code,
8757 so that it can be scanned to be kept in a safe place. This feature is
8758 particularly useful in combination with systemd-homed's support for
8759 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
8760 security tokens are lost. Recovery keys may be entered wherever the
8761 system asks for a password.
8763 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
8764 home directory which indicates that a home directory has not been
8765 deactivated cleanly when offline. This flag is useful to identify
8766 home directories for which the offline discard logic did not run when
8767 offlining, and where it would be a good idea to log in again to catch
8770 * systemctl gained a new parameter --timestamp= which may be used to
8771 change the style in which timestamps are output, i.e. whether to show
8772 them in local timezone or UTC, or whether to show µs granularity.
8774 * Alibaba's "pouch" container manager is now detected by
8775 systemd-detect-virt, ConditionVirtualization= and similar
8776 constructs. Similar, they now also recognize IBM PowerVM machine
8779 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
8780 place to use for propagating external mounts into the
8781 container. Similarly /run/host/notify is now used as the socket path
8782 for container payloads to communicate with the container manager
8783 using sd_notify(). The container manager now uses the
8784 /run/host/inaccessible/ directory to place "inaccessible" file nodes
8785 of all relevant types which may be used by the container payload as
8786 bind mount source to over-mount inodes to make them inaccessible.
8787 /run/host/container-manager will now be initialized with the same
8788 string as the $container environment variable passed to the
8789 container's PID 1. /run/host/container-uuid will be initialized with
8790 the same string as $container_uuid. This means the /run/host/
8791 hierarchy is now the primary way to make host resources available to
8792 the container. The Container Interface documents these new files and
8795 https://systemd.io/CONTAINER_INTERFACE
8797 * Support for the "ConditionNull=" unit file condition has been
8798 deprecated and undocumented for 6 years. systemd started to warn
8799 about its use 1.5 years ago. It has now been removed entirely.
8801 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
8802 a sd_bus_error struct and a list of error names, and checks if the
8803 error matches one of these names. It's a convenience wrapper that is
8804 useful in cases where multiple errors shall be handled the same way.
8806 * A new system call filter list "@known" has been added, that contains
8807 all system calls known at the time systemd was built.
8809 * Behaviour of system call filter allow lists has changed slightly:
8810 system calls that are contained in @known will result in EPERM by
8811 default, while those not contained in it result in ENOSYS. This
8812 should improve compatibility because known system calls will thus be
8813 communicated as prohibited, while unknown (and thus newer ones) will
8814 be communicated as not implemented, which hopefully has the greatest
8815 chance of triggering the right fallback code paths in client
8818 * "systemd-analyze syscall-filter" will now show two separate sections
8819 at the bottom of the output: system calls known during systemd build
8820 time but not included in any of the filter groups shown above, and
8821 system calls defined on the local kernel but known during systemd
8824 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
8825 systemd-nspawn all system call filter violations will be logged by
8826 the kernel (audit). This is useful for tracking down system calls
8827 invoked by container payloads that are prohibited by the container's
8828 system call filter policy.
8830 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
8831 systemd-nspawn (and other programs that use seccomp) all seccomp
8832 filtering is turned off.
8834 * Two new unit file settings ProtectProc= and ProcSubset= have been
8835 added that expose the hidepid= and subset= mount options of procfs.
8836 All processes of the unit will only see processes in /proc that are
8837 are owned by the unit's user. This is an important new sandboxing
8838 option that is recommended to be set on all system services. All
8839 long-running system services that are included in systemd itself set
8840 this option now. This option is only supported on kernel 5.8 and
8841 above, since the hidepid= option supported on older kernels was not a
8842 per-mount option but actually applied to the whole PID namespace.
8844 * Socket units gained a new boolean setting FlushPending=. If enabled
8845 all pending socket data/connections are flushed whenever the socket
8846 unit enters the "listening" state, i.e. after the associated service
8849 * The unit file setting NUMAMask= gained a new "all" value: when used,
8850 all existing NUMA nodes are added to the NUMA mask.
8852 * A new "credentials" logic has been added to system services. This is
8853 a simple mechanism to pass privileged data to services in a safe and
8854 secure way. It's supposed to be used to pass per-service secret data
8855 such as passwords or cryptographic keys but also associated less
8856 private information such as user names, certificates, and similar to
8857 system services. Each credential is identified by a short user-chosen
8858 name and may contain arbitrary binary data. Two new unit file
8859 settings have been added: SetCredential= and LoadCredential=. The
8860 former allows setting a credential to a literal string, the latter
8861 sets a credential to the contents of a file (or data read from a
8862 user-chosen AF_UNIX stream socket). Credentials are passed to the
8863 service via a special credentials directory, one file for each
8864 credential. The path to the credentials directory is passed in a new
8865 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
8866 are passed in the file system they may be easily referenced in
8867 ExecStart= command lines too, thus no explicit support for the
8868 credentials logic in daemons is required (though ideally daemons
8869 would look for the bits they need in $CREDENTIALS_DIRECTORY
8870 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
8871 backed by unswappable memory if privileges allow it, immutable if
8872 privileges allow it, is accessible only to the service's UID, and is
8873 automatically destroyed when the service stops.
8875 * systemd-nspawn supports the same credentials logic. It can both
8876 consume credentials passed to it via the aforementioned
8877 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
8878 to its payload. The service manager/PID 1 has been updated to match
8879 this: it can also accept credentials from the container manager that
8880 invokes it (in fact: any process that invokes it), and passes them on
8881 to its services. Thus, credentials can be propagated recursively down
8882 the tree: from a system's service manager to a systemd-nspawn
8883 service, to the service manager that runs as container payload and to
8884 the service it runs below. Credentials may also be added on the
8885 systemd-nspawn command line, using new --set-credential= and
8886 --load-credential= command line switches that match the
8887 aforementioned service settings.
8889 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
8890 the partition drop-ins which may be used to format/LUKS
8891 encrypt/populate any created partitions. The partitions are
8892 encrypted/formatted/populated before they are registered in the
8893 partition table, so that they appear atomically: either the
8894 partitions do not exist yet or they exist fully encrypted, formatted,
8895 and populated — there is no time window where they are
8896 "half-initialized". Thus the system is robust to abrupt shutdown: if
8897 the tool is terminated half-way during its operations on next boot it
8898 will start from the beginning.
8900 * systemd-repart's --size= operation gained a new "auto" value. If
8901 specified, and operating on a loopback file it is automatically sized
8902 to the minimal size the size constraints permit. This is useful to
8903 use "systemd-repart" as an image builder for minimally sized images.
8905 * systemd-resolved now gained a third IPC interface for requesting name
8906 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
8907 interface is now supported. The nss-resolve NSS module has been
8908 modified to use this new interface instead of D-Bus. Using Varlink
8909 has a major benefit over D-Bus: it works without a broker service,
8910 and thus already during earliest boot, before the dbus daemon has
8911 been started. This means name resolution via systemd-resolved now
8912 works at the same time systemd-networkd operates: from earliest boot
8913 on, including in the initrd.
8915 * systemd-resolved gained support for a new DNSStubListenerExtra=
8916 configuration file setting which may be used to specify additional IP
8917 addresses the built-in DNS stub shall listen on, in addition to the
8918 main one on 127.0.0.53:53.
8920 * Name lookups issued via systemd-resolved's D-Bus and Varlink
8921 interfaces (and thus also via glibc NSS if nss-resolve is used) will
8922 now honour a trailing dot in the hostname: if specified the search
8923 path logic is turned off. Thus "resolvectl query foo." is now
8924 equivalent to "resolvectl query --search=off foo.".
8926 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
8927 exposes how /etc/resolv.conf is currently managed: by resolved (and
8928 in which mode if so) or another subsystem. "resolvctl" will display
8929 this property in its status output.
8931 * The resolv.conf snippets systemd-resolved provides will now set "."
8932 as the search domain if no other search domain is known. This turns
8933 off the derivation of an implicit search domain by nss-dns for the
8934 hostname, when the hostname is set to an FQDN. This change is done to
8935 make nss-dns using resolv.conf provided by systemd-resolved behave
8936 more similarly to nss-resolve.
8938 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
8939 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
8940 "birth" time (btime) of a file in addition to the atime, mtime, and
8943 * systemd-analyze gained a new verb "capability" that lists all known
8944 capabilities by the systemd build and by the kernel.
8946 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
8947 advance the system clock to it at boot if it is noticed to be before
8948 that time. Previously, PID 1 would only advance the time to an epoch
8949 time that is set during build-time. With this new file OS builders
8950 can change this epoch timestamp on individual OS images without
8951 having to rebuild systemd.
8953 * systemd-logind will now listen to the KEY_RESTART key from the Linux
8954 input layer and reboot the system if it is pressed, similarly to how
8955 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
8956 was originally defined in the Multimedia context (to restart playback
8957 of a song or film), but is now primarily used in various embedded
8958 devices for "Reboot" buttons. Accordingly, systemd-logind will now
8959 honour it as such. This may configured in more detail via the new
8960 HandleRebootKey= and RebootKeyIgnoreInhibited=.
8962 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
8963 copying OS trees, for example in "systemd-nspawn --ephemeral",
8964 "systemd-nspawn --template=", "machinectl clone" and similar. This is
8965 useful when operating with OSTree images, which use hardlinks heavily
8966 throughout, and where such copies previously resulting in "exploding"
8969 * systemd-nspawn's --console= setting gained support for a new
8970 "autopipe" value, which is identical to "interactive" when invoked on
8971 a TTY, and "pipe" otherwise.
8973 * systemd-networkd's .network files gained support for explicitly
8974 configuring the multicast membership entries of bridge devices in the
8975 [BridgeMDB] section. It also gained support for the PIE queuing
8976 discipline in the [FlowQueuePIE] sections.
8978 * systemd-networkd's .netdev files may now be used to create "BareUDP"
8979 tunnels, configured in the new [BareUDP] setting.
8981 * systemd-networkd's Gateway= setting in .network files now accepts the
8982 special values "_dhcp4" and "_ipv6ra" to configure additional,
8983 locally defined, explicit routes to the gateway acquired via DHCP or
8984 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
8985 but still accepted for backwards compatibility.
8987 * systemd-networkd's [IPv6PrefixDelegation] section and
8988 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
8989 IPv6SendRA= (the old names are still accepted for backwards
8992 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
8993 boolean setting in [Network] section. If enabled, the delegated prefix
8994 gained by another link will be configured, and an address within the
8995 prefix will be assigned.
8997 * systemd-networkd's .network files gained the Announce= boolean setting
8998 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
8999 prefix will be announced through IPv6 router advertisement (IPv6 RA).
9000 The setting is enabled by default.
9002 * VXLAN tunnels may now be marked as independent of any underlying
9003 network interface via the new Independent= boolean setting.
9005 * systemctl gained support for two new verbs: "service-log-level" and
9006 "service-log-target" may be used on services that implement the
9007 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
9008 adjust the log level and target. All of systemd's long-running
9009 services support this now, but ideally all system services would
9010 implement this interface to make the system more uniformly
9013 * The SystemCallErrorNumber= unit file setting now accepts the new
9014 "kill" and "log" actions, in addition to arbitrary error number
9015 specifications as before. If "kill" the processes are killed on the
9016 event, if "log" the offending system call is audit logged.
9018 * A new SystemCallLog= unit file setting has been added that accepts a
9019 list of system calls that shall be logged about (audit).
9021 * The OS image dissection logic (as used by RootImage= in unit files or
9022 systemd-nspawn's --image= switch) has gained support for identifying
9023 and mounting explicit /usr/ partitions, which are now defined in the
9024 discoverable partition specification. This should be useful for
9025 environments where the root file system is
9026 generated/formatted/populated dynamically on first boot and combined
9027 with an immutable /usr/ tree that is supplied by the vendor.
9029 * In the final phase of shutdown, within the systemd-shutdown binary
9030 we'll now try to detach MD devices (i.e software RAID) in addition to
9031 loopback block devices and DM devices as before. This is supposed to
9032 be a safety net only, in order to increase robustness if things go
9033 wrong. Storage subsystems are expected to properly detach their
9034 storage volumes during regular shutdown already (or in case of
9035 storage backing the root file system: in the initrd hook we return to
9038 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
9039 will now log the thread ID in their log output. This is useful when
9040 working with heavily threaded programs.
9042 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
9043 not use the RDRAND CPU instruction. This is useful in environments
9044 such as replay debuggers where non-deterministic behaviour is not
9047 * The autopaging logic in systemd's various tools (such as systemctl)
9048 has been updated to turn on "secure" mode in "less"
9049 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
9050 detected. This disables invoking external programs from the pager,
9051 via the pipe logic. This behaviour may be overridden via the new
9052 $SYSTEMD_PAGERSECURE environment variable.
9054 * Units which have resource limits (.service, .mount, .swap, .slice,
9055 .socket, and .slice) gained new configuration settings
9056 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
9057 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
9058 limits and optional action taken by systemd-oomd.
9060 * A new service systemd-oomd has been added. It monitors resource
9061 contention for selected parts of the unit hierarchy using the PSI
9062 information reported by the kernel, and kills processes when memory
9063 or swap pressure is above configured limits. This service is only
9064 enabled by default in developer mode (see below) and should be
9065 considered a preview in this release. Behaviour details and option
9066 names are subject to change without the usual backwards-compatibility
9069 * A new helper oomctl has been added to introspect systemd-oomd state.
9070 It is only enabled by default in developer mode and should be
9071 considered a preview without the usual backwards-compatibility
9074 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
9075 enabled, systemd reads the system UID boundaries from /etc/login.defs
9076 at runtime, instead of using the built-in values selected during
9077 build. This is an option to improve compatibility for upgrades from
9078 old systems. It's strongly recommended not to make use of this
9079 functionality on new systems (or even enable it during build), as it
9080 makes something runtime-configurable that is mostly an implementation
9081 detail of the OS, and permits avoidable differences in deployments
9082 that create all kinds of problems in the long run.
9084 * New meson option '-Dmode=developer|release' has been added. When
9085 'developer', additional checks and features are enabled that are
9086 relevant during upstream development, e.g. verification that
9087 semi-automatically-generated documentation has been properly updated
9088 following API changes. Those checks are considered hints for
9089 developers and are not actionable in downstream builds. In addition,
9090 extra features that are not ready for general consumption may be
9091 enabled in developer mode. It is thus recommended to set
9092 '-Dmode=release' in end-user and distro builds.
9094 * systemd-cryptsetup gained support for processing detached LUKS
9095 headers specified on the kernel command line via the header=
9096 parameter of the luks.options= kernel command line option. The same
9097 device/path syntax as for key files is supported for header files
9100 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
9101 slot index data for devices that are connected through a PCI bridge
9102 where the _SUN index is associated with the bridge instead of the
9103 network device itself. Previously this would create ambiguous device
9104 naming if multiple network interfaces were connected to the same PCI
9105 bridge. Since this is a naming scheme incompatibility on systems that
9106 possess hardware like this it has been introduced as new naming
9107 scheme "v247". The previous scheme can be selected via the
9108 "net.naming_scheme=v245" kernel command line parameter.
9110 * ConditionFirstBoot= semantics have been modified to be safe towards
9111 abnormal system power-off during first boot. Specifically, the
9112 "systemd-machine-id-commit.service" service now acts as boot
9113 milestone indicating when the first boot process is sufficiently
9114 complete in order to not consider the next following boot also a
9115 first boot. If the system is reset before this unit is reached the
9116 first time, the next boot will still be considered a first boot; once
9117 it has been reached, no further boots will be considered a first
9118 boot. The "first-boot-complete.target" unit now acts as official hook
9119 point to order against this. If a service shall be run on every boot
9120 until the first boot fully succeeds it may thus be ordered before
9121 this target unit (and pull it in) and carry ConditionFirstBoot=
9124 * bootctl's set-default and set-oneshot commands now accept the three
9125 special strings "@default", "@oneshot", "@current" in place of a boot
9126 entry id. These strings are resolved to the current default and
9127 oneshot boot loader entry, as well as the currently booted one. Thus
9128 a command "bootctl set-default @current" may be used to make the
9129 currently boot menu item the new default for all subsequent boots.
9131 * "systemctl edit" has been updated to show the original effective unit
9132 contents in commented form in the text editor.
9134 * Units in user mode are now segregated into three new slices:
9135 session.slice (units that form the core of graphical session),
9136 app.slice ("normal" user applications), and background.slice
9137 (low-priority tasks). Unless otherwise configured, user units are
9138 placed in app.slice. The plan is to add resource limits and
9139 protections for the different slices in the future.
9141 * New GPT partition types for RISCV32/64 for the root and /usr
9142 partitions, and their associated Verity partitions have been defined,
9143 and are now understood by systemd-gpt-auto-generator, and the OS
9144 image dissection logic.
9146 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
9147 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
9148 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
9149 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
9150 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
9151 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
9152 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
9153 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
9154 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
9155 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
9156 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
9157 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
9158 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
9159 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
9160 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
9161 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
9162 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
9163 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
9164 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
9165 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
9166 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
9167 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
9168 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
9169 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
9170 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
9171 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
9172 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
9173 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
9174 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
9175 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
9176 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
9177 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
9178 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
9180 – Warsaw, 2020-11-26
9184 * The service manager gained basic support for cgroup v2 freezer. Units
9185 can now be suspended or resumed either using new systemctl verbs,
9186 freeze and thaw respectively, or via D-Bus.
9188 * PID 1 may now automatically load pre-compiled AppArmor policies from
9189 /etc/apparmor/earlypolicy during early boot.
9191 * The CPUAffinity= setting in service unit files now supports a new
9192 special value "numa" that causes the CPU affinity masked to be set
9193 based on the NUMA mask.
9195 * systemd will now log about all left-over processes remaining in a
9196 unit when the unit is stopped. It will now warn about services using
9197 KillMode=none, as this is generally an unsafe thing to make use of.
9199 * Two new unit file settings
9200 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
9201 added. They may be used to check whether a specific file system path
9202 resides on a block device that is encrypted on the block level
9203 (i.e. using dm-crypt/LUKS).
9205 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
9206 has been added that may be used for simple environment checks. This
9207 is particularly useful when passing in environment variables from a
9208 container manager (or from PAM in case of the systemd --user
9211 * .service unit files now accept a new setting CoredumpFilter= which
9212 allows configuration of the memory sections coredumps of the
9213 service's processes shall include.
9215 * .mount units gained a new ReadWriteOnly= boolean option. If set
9216 it will not be attempted to mount a file system read-only if mounting
9217 in read-write mode doesn't succeed. An option x-systemd.rw-only is
9218 available in /etc/fstab to control the same.
9220 * .socket units gained a new boolean setting PassPacketInfo=. If
9221 enabled, the kernel will attach additional per-packet metadata to all
9222 packets read from the socket, as an ancillary message. This controls
9223 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
9224 depending on socket type.
9226 * .service units gained a new setting RootHash= which may be used to
9227 specify the root hash for verity enabled disk images which are
9228 specified in RootImage=. RootVerity= may be used to specify a path to
9229 the Verity data matching a RootImage= file system. (The latter is
9230 only useful for images that do not contain the Verity data embedded
9231 into the same image that carries a GPT partition table following the
9232 Discoverable Partition Specification). Similarly, systemd-nspawn
9233 gained a new switch --verity-data= that takes a path to a file with
9234 the verity data of the disk image supplied in --image=, if the image
9235 doesn't contain the verity data itself.
9237 * .service units gained a new setting RootHashSignature= which takes
9238 either a base64 encoded PKCS#7 signature of the root hash specified
9239 with RootHash=, or a path to a file to read the signature from. This
9240 allows validation of the root hash against public keys available in
9241 the kernel keyring, and is only supported on recent kernels
9242 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
9243 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
9244 this mechanism has also been added to systemd-veritysetup.
9246 * .service unit files gained two new options
9247 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
9248 tune behaviour if a start or stop timeout is hit, i.e. whether to
9249 terminate the service with SIGTERM, SIGABRT or SIGKILL.
9251 * Most options in systemd that accept hexadecimal values prefixed with
9252 0x in additional to the usual decimal notation now also support octal
9253 notation when the 0o prefix is used and binary notation if the 0b
9256 * Various command line parameters and configuration file settings that
9257 configure key or certificate files now optionally take paths to
9258 AF_UNIX sockets in the file system. If configured that way a stream
9259 connection is made to the socket and the required data read from
9260 it. This is a simple and natural extension to the existing regular
9261 file logic, and permits other software to provide keys or
9262 certificates via simple IPC services, for example when unencrypted
9263 storage on disk is not desired. Specifically, systemd-networkd's
9264 Wireguard and MACSEC key file settings as well as
9265 systemd-journal-gatewayd's and systemd-journal-remote's PEM
9266 key/certificate parameters support this now.
9268 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
9269 configuration files that support specifier expansion learnt six new
9270 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
9271 resolve to the various ID fields from /etc/os-release, %l resolves to
9272 the "short" hostname of the system, i.e. the hostname configured in
9273 the kernel truncated at the first dot.
9275 * Support for the .include syntax in unit files has been removed. The
9276 concept has been obsolete for 6 years and we started warning about
9277 its pending removal 2 years ago (also see NEWS file below). It's
9280 * StandardError= and StandardOutput= in unit files no longer support
9281 the "syslog" and "syslog-console" switches. They were long removed
9282 from the documentation, but will now result in warnings when used,
9283 and be converted to "journal" and "journal+console" automatically.
9285 * If the service setting User= is set to the "nobody" user, a warning
9286 message is now written to the logs (but the value is nonetheless
9287 accepted). Setting User=nobody is unsafe, since the primary purpose
9288 of the "nobody" user is to own all files whose owner cannot be mapped
9289 locally. It's in particular used by the NFS subsystem and in user
9290 namespacing. By running a service under this user's UID it might get
9291 read and even write access to all these otherwise unmappable files,
9292 which is quite likely a major security problem.
9294 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
9295 and others) now have a size and inode limits applied (50% of RAM for
9296 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
9297 that the implicit kernel default is 50% too, so there is no change
9298 in the size limit for /tmp and /dev/shm.
9300 * nss-mymachines lost support for resolution of users and groups, and
9301 now only does resolution of hostnames. This functionality is now
9302 provided by nss-systemd. Thus, the 'mymachines' entry should be
9303 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
9304 (and 'systemd' added if it is not already there).
9306 * A new kernel command line option systemd.hostname= has been added
9307 that allows controlling the hostname that is initialized early during
9310 * A kernel command line option "udev.blockdev_read_only" has been
9311 added. If specified all hardware block devices that show up are
9312 immediately marked as read-only by udev. This option is useful for
9313 making sure that a specific boot under no circumstances modifies data
9314 on disk. Use "blockdev --setrw" to undo the effect of this, per
9317 * A new boolean kernel command line option systemd.swap= has been
9318 added, which may be used to turn off automatic activation of swap
9319 devices listed in /etc/fstab.
9321 * New kernel command line options systemd.condition_needs_update= and
9322 systemd.condition_first_boot= have been added, which override the
9323 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
9326 * A new kernel command line option systemd.clock_usec= has been added
9327 that allows setting the system clock to the specified time in µs
9328 since Jan 1st, 1970 early during boot. This is in particular useful
9329 in order to make test cases more reliable.
9331 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
9332 systemd-coredump to save core files for suid processes. When saving
9333 the core file, systemd-coredump will use the effective uid and gid of
9334 the process that faulted.
9336 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
9337 now automatically set to "Y" at boot, in order to enable pstore
9338 generation for collection with systemd-pstore.
9340 * We provide a set of udev rules to enable auto-suspend on PCI and USB
9341 devices that were tested to correctly support it. Previously, this
9342 was distributed as a set of udev rules, but has now been replaced by
9343 by a set of hwdb entries (and a much shorter udev rule to take action
9344 if the device modalias matches one of the new hwdb entries).
9346 As before, entries are periodically imported from the database
9347 maintained by the ChromiumOS project. If you have a device that
9348 supports auto-suspend correctly and where it should be enabled by
9349 default, please submit a patch that adds it to the database (see
9350 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
9352 * systemd-udevd gained the new configuration option timeout_signal= as well
9353 as a corresponding kernel command line option udev.timeout_signal=.
9354 The option can be used to configure the UNIX signal that the main
9355 daemon sends to the worker processes on timeout. Setting the signal
9356 to SIGABRT is useful for debugging.
9358 * .link files managed by systemd-udevd gained options RxFlowControl=,
9359 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
9360 order to configure various flow control parameters. They also gained
9361 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
9362 frame ring buffer sizes.
9364 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
9365 enabled systemd-networkd manages all routes configured by other tools.
9367 * .network files managed by systemd-networkd gained a new section
9368 [SR-IOV], in order to configure SR-IOV capable network devices.
9370 * systemd-networkd's [IPv6Prefix] section in .network files gained a
9371 new boolean setting Assign=. If enabled an address from the prefix is
9372 automatically assigned to the interface.
9374 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
9375 controls delegated prefixes assigned by DHCPv6 client. The section
9376 has three settings: SubnetID=, Assign=, and Token=. The setting
9377 SubnetID= allows explicit configuration of the preferred subnet that
9378 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
9379 Assign= is enabled (which is the default) an address from any acquired
9380 delegated prefix is automatically chosen and assigned to the
9381 interface. The setting Token= specifies an optional address generation
9384 * systemd-networkd's [Network] section gained a new setting
9385 IPv4AcceptLocal=. If enabled the interface accepts packets with local
9388 * systemd-networkd gained support for configuring the HTB queuing
9389 discipline in the [HierarchyTokenBucket] and
9390 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
9391 be configured in the [PFIFO] section, "GRED" in
9392 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
9393 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
9394 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
9395 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
9396 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
9397 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
9399 * systemd-networkd gained support for a new Termination= setting in the
9400 [CAN] section for configuring the termination resistor. It also
9401 gained a new ListenOnly= setting for controlling whether to only
9402 listen on CAN interfaces, without interfering with traffic otherwise
9403 (which is useful for debugging/monitoring CAN network
9404 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
9405 been added to configure various CAN-FD aspects.
9407 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
9408 When enabled, DHCPv6 will be attempted right-away without requiring an
9409 Router Advertisement packet suggesting it first (i.e. without the 'M'
9410 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
9411 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
9412 the RA packets suggest it.
9414 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
9415 which may be used to turn off use of the gateway information provided
9416 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
9417 used to configure how to process leases that lack a lifetime option.
9419 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
9420 setting SendVendorOption= allowing configuration of additional vendor
9421 options to send in the DHCP requests/responses. The [DHCPv6] section
9422 gained a new SendOption= setting for sending arbitrary DHCP
9423 options. RequestOptions= has been added to request arbitrary options
9424 from the server. UserClass= has been added to set the DHCP user class
9427 * systemd-networkd's [DHCPServer] section gained a new set of options
9428 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
9429 information about these three protocols in the DHCP lease. It also
9430 gained support for including "MUD" URLs ("Manufacturer Usage
9431 Description"). Support for "MUD" URLs was also added to the LLDP
9432 stack, configurable in the [LLDP] section in .network files.
9434 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
9435 mode. Also, the sections now support a new setting SourceMACAddress=.
9437 * systemd-networkd's .netdev files now support a new setting
9438 VLANProtocol= in the [Bridge] section that allows configuration of
9439 the VLAN protocol to use.
9441 * systemd-networkd supports a new Group= setting in the [Link] section
9442 of the .network files, to control the link group.
9444 * systemd-networkd's [Network] section gained a new
9445 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
9446 link local address is generated.
9448 * A new default .network file is now shipped that matches TUN/TAP
9449 devices that begin with "vt-" in their name. Such interfaces will
9450 have IP routing onto the host links set up automatically. This is
9451 supposed to be used by VM managers to trivially acquire a network
9452 interface which is fully set up for host communication, simply by
9453 carefully picking an interface name to use.
9455 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
9456 which sets the route priority for routes specified by the DHCP server.
9458 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
9459 which configures the vendor class information sent to DHCP server.
9461 * The BlackList= settings in .network files' [DHCPv4] and
9462 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
9463 are still understood to provide compatibility.
9465 * networkctl gained the new "forcerenew" command for forcing all DHCP
9466 server clients to renew their lease. The interface "status" output
9467 will now show numerous additional fields of information about an
9468 interface. There are new "up" and "down" commands to bring specific
9469 interfaces up or down.
9471 * systemd-resolved's DNS= configuration option now optionally accepts a
9472 port number (after ":") and a host name (after "#"). When the host
9473 name is specified, the DNS-over-TLS certificate is validated to match
9474 the specified hostname. Additionally, in case of IPv6 addresses, an
9475 interface may be specified (after "%").
9477 * systemd-resolved may be configured to forward single-label DNS names.
9478 This is not standard-conformant, but may make sense in setups where
9479 public DNS servers are not used.
9481 * systemd-resolved's DNS-over-TLS support gained SNI validation.
9483 * systemd-nspawn's --resolv-conf= switch gained a number of new
9484 supported values. Specifically, options starting with "replace-" are
9485 like those prefixed "copy-" but replace any existing resolv.conf
9486 file. And options ending in "-uplink" and "-stub" can now be used to
9487 propagate other flavours of resolv.conf into the container (as
9488 defined by systemd-resolved).
9490 * The various programs included in systemd can now optionally output
9491 their log messages on stderr prefixed with a timestamp, controlled by
9492 the $SYSTEMD_LOG_TIME environment variable.
9494 * systemctl gained a new "-P" switch that is a shortcut for "--value
9497 * "systemctl list-units" and "systemctl list-machines" no longer hide
9498 their first output column with --no-legend. To hide the first column,
9501 * "systemctl reboot" takes the option "--reboot-argument=".
9502 The optional positional argument to "systemctl reboot" is now
9503 being deprecated in favor of this option.
9505 * systemd-run gained a new switch --slice-inherit. If specified the
9506 unit it generates is placed in the same slice as the systemd-run
9509 * systemd-journald gained support for zstd compression of large fields
9510 in journal files. The hash tables in journal files have been hardened
9511 against hash collisions. This is an incompatible change and means
9512 that journal files created with new systemd versions are not readable
9513 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
9514 environment variable for systemd-journald.service is set to 0 this
9515 new hardening functionality may be turned off, so that generated
9516 journal files remain compatible with older journalctl
9519 * journalctl will now include a clickable link in the default output for
9520 each log message for which a URL with further documentation is
9521 known. This is only supported on terminal emulators that support
9522 clickable hyperlinks, and is turned off if a pager is used (since
9523 "less" still doesn't support hyperlinks,
9524 unfortunately). Documentation URLs may be included in log messages
9525 either by including a DOCUMENTATION= journal field in it, or by
9526 associating a journal message catalog entry with the log message's
9527 MESSAGE_ID, which then carries a "Documentation:" tag.
9529 * journald.conf gained a new boolean setting Audit= that may be used to
9530 control whether systemd-journald will enable audit during
9533 * when systemd-journald's log stream is broken up into multiple lines
9534 because the PID of the sender changed this is indicated in the
9535 generated log records via the _LINE_BREAK=pid-change field.
9537 * journalctl's "-o cat" output mode will now show one or more journal
9538 fields specified with --output-fields= instead of unconditionally
9539 MESSAGE=. This is useful to retrieve a very specific set of fields
9540 without any decoration.
9542 * The sd-journal.h API gained two new functions:
9543 sd_journal_enumerate_available_unique() and
9544 sd_journal_enumerate_available_data() that operate like their
9545 counterparts that lack the _available_ in the name, but skip items
9546 that cannot be read and processed by the local implementation
9547 (i.e. are compressed in an unsupported format or such),
9549 * coredumpctl gained a new --file= switch, matching the same one in
9550 journalctl: a specific journal file may be specified to read the
9553 * coredumps collected by systemd-coredump may now be compressed using
9556 * systemd-binfmt gained a new switch --unregister for unregistering all
9557 registered entries at once. This is now invoked automatically at
9558 shutdown, so that binary formats registered with the "F" flag will
9559 not block clean file system unmounting.
9561 * systemd-notify's --pid= switch gained new values: "parent", "self",
9562 "auto" for controlling which PID to send to the service manager: the
9563 systemd-notify process' PID, or the one of the process invoking it.
9565 * systemd-logind's Session bus object learnt a new method call
9566 SetType() for temporarily updating the session type of an already
9567 allocated session. This is useful for upgrading tty sessions to
9568 graphical ones once a compositor is invoked.
9570 * systemd-socket-proxy gained a new switch --exit-idle-time= for
9571 configuring an exit-on-idle time.
9573 * systemd-repart's --empty= setting gained a new value "create". If
9574 specified a new empty regular disk image file is created under the
9575 specified name. Its size may be specified with the new --size=
9576 option. The latter is also supported without the "create" mode, in
9577 order to grow existing disk image files to the specified size. These
9578 two new options are useful when creating or manipulating disk images
9579 instead of operating on actual block devices.
9581 * systemd-repart drop-ins now support a new UUID= setting to control
9582 the UUID to assign to a newly created partition.
9584 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
9587 * systemd-repart's Label= setting now support the usual, simple
9588 specifier expansion.
9590 * systemd-homed's LUKS backend gained the ability to discard empty file
9591 system blocks automatically when the user logs out. This is enabled
9592 by default to ensure that home directories take minimal space when
9593 logged out but get full size guarantees when logged in. This may be
9594 controlled with the new --luks-offline-discard= switch to homectl.
9596 * If systemd-homed detects that /home/ is encrypted as a whole it will
9597 now default to the directory or subvolume backends instead of the
9598 LUKS backend, in order to avoid double encryption. The default
9599 storage and file system may now be configured explicitly, too, via
9600 the new /etc/systemd/homed.conf configuration file.
9602 * systemd-homed now supports unlocking home directories with FIDO2
9603 security tokens that support the 'hmac-secret' extension, in addition
9604 to the existing support for PKCS#11 security token unlocking
9605 support. Note that many recent hardware security tokens support both
9606 interfaces. The FIDO2 support is accessible via homectl's
9607 --fido2-device= option.
9609 * homectl's --pkcs11-uri= setting now accepts two special parameters:
9610 if "auto" is specified and only one suitable PKCS#11 security token
9611 is plugged in, its URL is automatically determined and enrolled for
9612 unlocking the home directory. If "list" is specified a brief table of
9613 suitable PKCS#11 security tokens is shown. Similar, the new
9614 --fido2-device= option also supports these two special values, for
9615 automatically selecting and listing suitable FIDO2 devices.
9617 * The /etc/crypttab tmp option now optionally takes an argument
9618 selecting the file system to use. Moreover, the default is now
9619 changed from ext2 to ext4.
9621 * There's a new /etc/crypttab option "keyfile-erase". If specified the
9622 key file listed in the same line is removed after use, regardless if
9623 volume activation was successful or not. This is useful if the key
9624 file is only acquired transiently at runtime and shall be erased
9625 before the system continues to boot.
9627 * There's also a new /etc/crypttab option "try-empty-password". If
9628 specified, before asking the user for a password it is attempted to
9629 unlock the volume with an empty password. This is useful for
9630 installing encrypted images whose password shall be set on first boot
9631 instead of at installation time.
9633 * systemd-cryptsetup will now attempt to load the keys to unlock
9634 volumes with automatically from files in
9635 /etc/cryptsetup-keys.d/<volume>.key and
9636 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
9638 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
9639 /etc/crypttab, during boot.
9641 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
9642 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
9645 * A new generator systemd-xdg-autostart-generator has been added. It
9646 generates systemd unit files from XDG autostart .desktop files, and
9647 may be used to let the systemd user instance manage services that are
9648 started automatically as part of the desktop session.
9650 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
9651 to query and change the firmware's 'Reboot Into Firmware Interface'
9654 * systemd-firstboot gained a new switch --kernel-command-line= that may
9655 be used to initialize the /etc/kernel/cmdline file of the image. It
9656 also gained a new switch --root-password-hashed= which is like
9657 --root-password= but accepts a pre-hashed UNIX password as
9658 argument. The new option --delete-root-password may be used to unset
9659 any password for the root user (dangerous!). The --root-shell= switch
9660 may be used to control the shell to use for the root account. A new
9661 --force option may be used to override any already set settings with
9662 the parameters specified on the command line (by default, the tool
9663 will not override what has already been set before, i.e. is purely
9666 * systemd-firstboot gained support for a new --image= switch, which is
9667 similar to --root= but accepts the path to a disk image file, on
9668 which it then operates.
9670 * A new sd-path.h API has been added to libsystemd. It provides a
9671 simple API for retrieving various search paths and primary
9672 directories for various resources.
9674 * A new call sd_notify_barrier() has been added to the sd-daemon.h
9675 API. The call will block until all previously sent sd_notify()
9676 messages have been processed by the service manager. This is useful
9677 to remove races caused by a process already having disappeared at the
9678 time a notification message is processed by the service manager,
9679 making correct attribution impossible. The systemd-notify tool will
9680 now make use of this call implicitly, but this can be turned off again
9681 via the new --no-block switch.
9683 * When sending a file descriptor (fd) to the service manager to keep
9684 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
9685 may be specified. If passed the service manager will refrain from
9686 poll()ing on the file descriptor. Traditionally (and when the
9687 parameter is not specified), the service manager will poll it for
9688 POLLHUP or POLLERR events, and immediately close the fds in that
9691 * The service manager (PID1) gained a new D-Bus method call
9692 SetShowStatus() which may be used to control whether it shall show
9693 boot-time status output on the console. This method has a similar
9694 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
9696 * The sd-bus API gained a number of convenience functions that take
9697 va_list arguments rather than "...". For example, there's now
9698 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
9699 it easier to build wrappers that accept variadic arguments and want
9700 to pass a ready va_list structure to sd-bus.
9702 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
9703 flag which alters how the userdata pointer to pass to the callbacks
9704 is determined. When the flag is set, the offset field is converted
9705 as-is into a pointer, without adding it to the object pointer the
9706 vtable is associated with.
9708 * sd-bus now exposes four new functions:
9709 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
9710 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
9711 validate strings to check if they qualify as various D-Bus concepts.
9713 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
9714 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
9715 that simplify adding argument names to D-Bus methods and signals.
9717 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
9719 * Various D-Bus APIs of systemd daemons now have man pages that
9720 document the methods, signals and properties.
9722 * The expectations on user/group name syntax are now documented in
9723 detail; documentation on how classic home directories may be
9724 converted into home directories managed by homed has been added;
9725 documentation regarding integration of homed/userdb functionality in
9726 desktops has been added:
9728 https://systemd.io/USER_NAMES
9729 https://systemd.io/CONVERTING_TO_HOMED
9730 https://systemd.io/USERDB_AND_DESKTOPS
9732 * Documentation for the on-disk Journal file format has been updated
9733 and has now moved to:
9735 https://systemd.io/JOURNAL_FILE_FORMAT
9737 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
9738 has been extended by a set of environment variables that expose
9739 select fields from the host's os-release file to the container
9740 payload. Similarly, host's os-release files can be mounted into the
9741 container underneath /run/host. Together, those mechanisms provide a
9742 standardized way to expose information about the host to the
9743 container payload. Both interfaces are implemented in systemd-nspawn.
9745 * All D-Bus services shipped in systemd now implement the generic
9746 LogControl1 D-Bus API which allows clients to change log level +
9747 target of the service during runtime.
9749 * Only relevant for developers: the mkosi.default symlink has been
9750 dropped from version control. Please create a symlink to one of the
9751 distribution-specific defaults in .mkosi/ based on your preference.
9753 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
9754 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
9755 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
9756 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
9757 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
9758 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
9759 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
9760 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
9761 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
9762 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
9763 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
9764 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
9765 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
9766 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
9767 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
9768 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
9769 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
9770 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
9771 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
9772 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
9773 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
9774 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
9775 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
9776 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
9777 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
9778 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
9779 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
9780 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
9781 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
9782 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
9783 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
9784 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
9785 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
9786 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
9787 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
9788 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
9789 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
9790 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
9792 – Warsaw, 2020-07-30
9796 * A new tool "systemd-repart" has been added, that operates as an
9797 idempotent declarative repartitioner for GPT partition tables.
9798 Specifically, a set of partitions that must or may exist can be
9799 configured via drop-in files, and during every boot the partition
9800 table on disk is compared with these files, creating missing
9801 partitions or growing existing ones based on configurable relative
9802 and absolute size constraints. The tool is strictly incremental,
9803 i.e. does not delete, shrink or move partitions, but only adds and
9804 grows them. The primary use-case is OS images that ship in minimized
9805 form, that on first boot are grown to the size of the underlying
9806 block device or augmented with additional partitions. For example,
9807 the root partition could be extended to cover the whole disk, or a
9808 swap or /home partitions could be added on first boot. It can also be
9809 used for systems that use an A/B update scheme but ship images with
9810 just the A partition, with B added on first boot. The tool is
9811 primarily intended to be run in the initrd, shortly before
9812 transitioning into the host OS, but can also be run after the
9813 transition took place. It automatically discovers the disk backing
9814 the root file system, and should hence not require any additional
9815 configuration besides the partition definition drop-ins. If no
9816 configuration drop-ins are present, no action is taken.
9818 * A new component "userdb" has been added, along with a small daemon
9819 "systemd-userdbd.service" and a client tool "userdbctl". The framework
9820 allows defining rich user and group records in a JSON format,
9821 extending on the classic "struct passwd" and "struct group"
9822 structures. Various components in systemd have been updated to
9823 process records in this format, including systemd-logind and
9824 pam-systemd. The user records are intended to be extensible, and
9825 allow setting various resource management, security and runtime
9826 parameters that shall be applied to processes and sessions of the
9827 user as they log in. This facility is intended to allow associating
9828 such metadata directly with user/group records so that they can be
9829 produced, extended and consumed in unified form. We hope that
9830 eventually frameworks such as sssd will generate records this way, so
9831 that for the first time resource management and various other
9832 per-user settings can be configured in LDAP directories and then
9833 provided to systemd (specifically to systemd-logind and pam-system)
9834 to apply on login. For further details see:
9836 https://systemd.io/USER_RECORD
9837 https://systemd.io/GROUP_RECORD
9838 https://systemd.io/USER_GROUP_API
9840 * A small new service systemd-homed.service has been added, that may be
9841 used to securely manage home directories with built-in encryption.
9842 The complete user record data is unified with the home directory,
9843 thus making home directories naturally migratable. Its primary
9844 back-end is based on LUKS volumes, but fscrypt, plain directories,
9845 and other storage schemes are also supported. This solves a couple of
9846 problems we saw with traditional ways to manage home directories, in
9847 particular when it comes to encryption. For further discussion of
9848 this, see the video of Lennart's talk at AllSystemsGo! 2019:
9850 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
9852 For further details about the format and expectations on home
9853 directories this new daemon makes, see:
9855 https://systemd.io/HOME_DIRECTORY
9857 * systemd-journald is now multi-instantiable. In addition to the main
9858 instance systemd-journald.service there's now a template unit
9859 systemd-journald@.service, with each instance defining a new named
9860 log 'namespace' (whose name is specified via the instance part of the
9861 unit name). A new unit file setting LogNamespace= has been added,
9862 taking such a namespace name, that assigns services to the specified
9863 log namespaces. As each log namespace is serviced by its own
9864 independent journal daemon, this functionality may be used to improve
9865 performance and increase isolation of applications, at the price of
9866 losing global message ordering. Each instance of journald has a
9867 separate set of configuration files, with possibly different disk
9868 usage limitations and other settings.
9870 journalctl now takes a new option --namespace= to show logs from a
9871 specific log namespace. The sd-journal.h API gained
9872 sd_journal_open_namespace() for opening the log stream of a specific
9873 log namespace. systemd-journald also gained the ability to exit on
9874 idle, which is useful in the context of log namespaces, as this means
9875 log daemons for log namespaces can be activated automatically on
9876 demand and will stop automatically when no longer used, minimizing
9879 * When systemd-tmpfiles copies a file tree using the 'C' line type it
9880 will now label every copied file according to the SELinux database.
9882 * When systemd/PID 1 detects it is used in the initrd it will now boot
9883 into initrd.target rather than default.target by default. This should
9884 make it simpler to build initrds with systemd as for many cases the
9885 only difference between a host OS image and an initrd image now is
9886 the presence of the /etc/initrd-release file.
9888 * A new kernel command line option systemd.cpu_affinity= is now
9889 understood. It's equivalent to the CPUAffinity= option in
9890 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
9891 itself and the default for all other processes.
9893 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
9894 equivalent), the SELinux database is now reloaded, ensuring that
9895 sockets and other file system objects are generated taking the new
9896 database into account.
9898 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
9899 "quiet" has been changed to imply that instead of
9900 "systemd.show-status=auto". In this mode, only messages about errors
9901 and significant delays in boot are shown on the console.
9903 * The sd-event.h API gained native support for the new Linux "pidfd"
9904 concept. This permits watching processes using file descriptors
9905 instead of PID numbers, which fixes a number of races and makes
9906 process supervision more robust and efficient. All of systemd's
9907 components will now use pidfds if the kernel supports it for process
9908 watching, with the exception of PID 1 itself, unfortunately. We hope
9909 to move PID 1 to exclusively using pidfds too eventually, but this
9910 requires some more kernel work first. (Background: PID 1 watches
9911 processes using waitid() with the P_ALL flag, and that does not play
9912 together nicely with pidfds yet.)
9914 * Closely related to this, the sd-event.h API gained two new calls
9915 sd_event_source_send_child_signal() (for sending a signal to a
9916 watched process) and sd_event_source_get_child_process_own() (for
9917 marking a process so that it is killed automatically whenever the
9918 event source watching it is freed).
9920 * systemd-networkd gained support for configuring Token Bucket Filter
9921 (TBF) parameters in its qdisc configuration support. Similarly,
9922 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
9923 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
9925 * systemd-networkd gained support for Intermediate Functional Block
9926 (IFB) network devices.
9928 * systemd-networkd gained support for configuring multi-path IP routes,
9929 using the new MultiPathRoute= setting in the [Route] section.
9931 * systemd-networkd's DHCPv4 client has been updated to support a new
9932 SendDecline= option. If enabled, duplicate address detection is done
9933 after a DHCP offer is received from the server. If a conflict is
9934 detected, the address is declined. The DHCPv4 client also gained
9935 support for a new RouteMTUBytes= setting that allows to configure the
9936 MTU size to be used for routes generated from DHCPv4 leases.
9938 * The PrefixRoute= setting in systemd-networkd's [Address] section of
9939 .network files has been deprecated, and replaced by AddPrefixRoute=,
9940 with its sense inverted.
9942 * The Gateway= setting of [Route] sections of .network files gained
9943 support for a special new value "_dhcp". If set, the configured
9944 static route uses the gateway host configured via DHCP.
9946 * New User= and SuppressPrefixLength= settings have been implemented
9947 for the [RoutingPolicyRule] section of .network files to configure
9948 source routing based on UID ranges and prefix length, respectively.
9950 * The Type= match property of .link files has been generalized to
9951 always match the device type shown by 'networkctl status', even for
9952 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
9955 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
9956 D-Bus message object as "sensitive". Those objects are erased from
9957 memory when they are freed. This concept is intended to be used for
9958 messages that contain security sensitive data. A new flag
9959 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
9960 in sd-bus vtables, causing any incoming and outgoing messages of
9961 those methods to be implicitly marked as "sensitive".
9963 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
9964 contents of a message (or parts thereof) to standard output for
9967 * systemd-sysusers gained support for creating users with the primary
9968 group named differently than the user.
9970 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
9971 gained support for growing XFS partitions. Previously it supported
9972 only ext4 and btrfs partitions.
9974 * The support for /etc/crypttab gained a new x-initrd.attach option. If
9975 set, the specified encrypted volume is unlocked already in the
9976 initrd. This concept corresponds to the x-initrd.mount option in
9979 * systemd-cryptsetup gained native support for unlocking encrypted
9980 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
9981 encryption of volumes to YubiKeys. This is exposed in the new
9982 pkcs11-uri= option in /etc/crypttab.
9984 * The /etc/fstab support in systemd now supports two new mount options
9985 x-systemd.{required,wanted}-by=, for explicitly configuring the units
9986 that the specified mount shall be pulled in by, in place of
9987 the usual local-fs.target/remote-fs.target.
9989 * The https://systemd.io/ web site has been relaunched, directly
9990 populated with most of the documentation included in the systemd
9991 repository. systemd also acquired a new logo, thanks to Tobias
9994 * systemd-udevd gained support for managing "alternative" network
9995 interface names, as supported by new Linux kernels. For the first
9996 time this permits assigning multiple (and longer!) names to a network
9997 interface. systemd-udevd will now by default assign the names
9998 generated via all supported naming schemes to each interface. This
9999 may be further tweaked with .link files and the AlternativeName= and
10000 AlternativeNamesPolicy= settings. Other components of systemd have
10001 been updated to support the new alternative names wherever
10002 appropriate. For example, systemd-nspawn will now generate
10003 alternative interface names for the host-facing side of container
10004 veth links based on the full container name without truncation.
10006 * systemd-nspawn interface naming logic has been updated in another way
10007 too: if the main interface name (i.e. as opposed to new-style
10008 "alternative" names) based on the container name is truncated, a
10009 simple hashing scheme is used to give different interface names to
10010 multiple containers whose names all begin with the same prefix. Since
10011 this changes the primary interface names pointing to containers if
10012 truncation happens, the old scheme may still be requested by
10013 selecting an older naming scheme, via the net.naming_scheme= kernel
10014 command line option.
10016 * PrivateUsers= in service files now works in services run by the
10017 systemd --user per-user instance of the service manager.
10019 * A new per-service sandboxing option ProtectClock= has been added that
10020 locks down write access to the system clock. It takes away device
10021 node access to /dev/rtc as well as the system calls that set the
10022 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
10023 Note that this option does not affect access to auxiliary services
10024 that allow changing the clock, for example access to
10027 * The systemd-id128 tool gained a new "show" verb for listing or
10028 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
10029 GPT partition table types.
10031 * The Discoverable Partitions Specification has been updated to support
10032 /var and /var/tmp partition discovery. Support for this has been
10033 added to systemd-gpt-auto-generator. For details see:
10035 https://systemd.io/DISCOVERABLE_PARTITIONS
10037 * "systemctl list-unit-files" has been updated to show a new column
10038 with the suggested enablement state based on the vendor preset files
10039 for the respective units.
10041 * "systemctl" gained a new option "--with-dependencies". If specified
10042 commands such as "systemctl status" or "systemctl cat" will now show
10043 all specified units along with all units they depend on.
10045 * networkctl gained support for showing per-interface logs in its
10048 * systemd-networkd-wait-online gained support for specifying the maximum
10049 operational state to wait for, and to wait for interfaces to
10052 * The [Match] section of .link and .network files now supports a new
10053 option PermanentMACAddress= which may be used to check against the
10054 permanent MAC address of a network device even if a randomized MAC
10057 * The [TrafficControlQueueingDiscipline] section in .network files has
10058 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
10059 dropped from the individual setting names.
10061 * Any .link and .network files that have an empty [Match] section (this
10062 also includes empty and commented-out files) will now be
10063 rejected. systemd-udev and systemd-networkd started warning about
10064 such files in version 243.
10066 * systemd-logind will now validate access to the operation of changing
10067 the virtual terminal via a polkit action. By default, only users
10068 with at least one session on a local VT are granted permission.
10070 * When systemd sets up PAM sessions that invoked service processes
10071 shall run in, the pam_setcred() API is now invoked, thus permitting
10072 PAM modules to set additional credentials for the processes.
10074 * portablectl attach/detach verbs now accept --now and --enable options
10075 to combine attachment with enablement and invocation, or detachment
10076 with stopping and disablement.
10078 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
10079 fixed, which in turn exposed bugs in unit configuration of services
10080 which have Type=oneshot and should only run once, but do not have
10081 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
10082 service may be started again after exiting successfully, for example
10083 as a dependency in another transaction. Affected services included
10084 some internal systemd services (most notably
10085 systemd-vconsole-setup.service, which was updated to have
10086 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
10087 plymouth has been suitably updated or patched before upgrading to
10088 this systemd release. See
10089 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
10090 additional discussion.
10092 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
10093 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
10094 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
10095 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
10096 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
10097 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
10098 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
10099 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
10100 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
10101 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
10102 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
10103 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
10104 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
10105 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
10106 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
10107 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
10108 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
10109 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
10110 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
10111 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
10112 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
10113 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
10114 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
10115 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
10118 – Warsaw, 2020-03-06
10122 * Support for the cpuset cgroups v2 controller has been added.
10123 Processes may be restricted to specific CPUs using the new
10124 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
10125 AllowedMemoryNodes= setting.
10127 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
10128 now be configured using a new RestartKillSignal= setting. This
10129 allows units which signals to request termination to implement
10130 different behaviour when stopping in preparation for a restart.
10132 * "systemctl clean" may now be used also for socket, mount, and swap
10135 * systemd will also read configuration options from the EFI variable
10136 SystemdOptions. This may be used to configure systemd behaviour when
10137 modifying the kernel command line is inconvenient, but configuration
10138 on disk is read too late, for example for the options related to
10139 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
10140 set the EFI variable.
10142 * systemd will now disable printk ratelimits in early boot. This should
10143 allow us to capture more logs from the early boot phase where normal
10144 storage is not available and the kernel ring buffer is used for
10145 logging. Configuration on the kernel command line has higher priority
10146 and overrides the systemd setting.
10148 systemd programs which log to /dev/kmsg directly use internal
10149 ratelimits to prevent runaway logging. (Normally this is only used
10150 during early boot, so in practice this change has very little
10153 * Unit files now support top level dropin directories of the form
10154 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
10155 that affects all corresponding unit files.
10157 * systemctl gained support for 'stop --job-mode=triggering' which will
10158 stop the specified unit and any units which could trigger it.
10160 * Unit status display now includes units triggering and triggered by
10161 the unit being shown.
10163 * The RuntimeMaxSec= setting is now supported by scopes, not just
10164 .service units. This is particularly useful for PAM sessions which
10165 create a scope unit for the user login. systemd.runtime_max_sec=
10166 setting may used with the pam_systemd module to limit the duration
10167 of the PAM session, for example for time-limited logins.
10169 * A new @pkey system call group is now defined to make it easier to
10170 allow-list memory protection syscalls for containers and services
10171 which need to use them.
10173 * systemd-udevd: removed the 30s timeout for killing stale workers on
10174 exit. systemd-udevd now waits for workers to finish. The hard-coded
10175 exit timeout of 30s was too short for some large installations, where
10176 driver initialization could be prematurely interrupted during initrd
10177 processing if the root file system had been mounted and init was
10178 preparing to switch root. If udevd is run without systemd and workers
10179 are hanging while udevd receives an exit signal, udevd will now exit
10180 when udev.event_timeout is reached for the last hanging worker. With
10181 systemd, the exit timeout can additionally be configured using
10182 TimeoutStopSec= in systemd-udevd.service.
10184 * udev now provides a program (fido_id) that identifies FIDO CTAP1
10185 ("U2F")/CTAP2 security tokens based on the usage declared in their
10186 report and descriptor and outputs suitable environment variables.
10187 This replaces the externally maintained allow lists of all known
10188 security tokens that were used previously.
10190 * Automatically generated autosuspend udev rules for allow-listed
10191 devices have been imported from the Chromium OS project. This should
10192 improve power saving with many more devices.
10194 * udev gained a new "CONST{key}=value" setting that allows matching
10195 against system-wide constants without forking a helper binary.
10196 Currently "arch" and "virt" keys are supported.
10198 * udev now opens CDROMs in non-exclusive mode when querying their
10199 capabilities. This should fix issues where other programs trying to
10200 use the CDROM cannot gain access to it, but carries a risk of
10201 interfering with programs writing to the disk, if they did not open
10202 the device in exclusive mode as they should.
10204 * systemd-networkd does not create a default route for IPv4 link local
10205 addressing anymore. The creation of the route was unexpected and was
10206 breaking routing in various cases, but people who rely on it being
10207 created implicitly will need to adjust. Such a route may be requested
10208 with DefaultRouteOnDevice=yes.
10210 Similarly, systemd-networkd will not assign a link-local IPv6 address
10211 when IPv6 link-local routing is not enabled.
10213 * Receive and transmit buffers may now be configured on links with
10214 the new RxBufferSize= and TxBufferSize= settings.
10216 * systemd-networkd may now advertise additional IPv6 routes. A new
10217 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
10220 * systemd-networkd may now configure "next hop" routes using the
10221 [NextHop] section and Gateway= and Id= settings.
10223 * systemd-networkd will now retain DHCP config on restarts by default
10224 (but this may be overridden using the KeepConfiguration= setting).
10225 The default for SendRelease= has been changed to true.
10227 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
10228 received from the server.
10230 The client will use the received SIP server list if UseSIP=yes is
10233 The client may be configured to request specific options from the
10234 server using a new RequestOptions= setting.
10236 The client may be configured to send arbitrary options to the server
10237 using a new SendOption= setting.
10239 A new IPServiceType= setting has been added to configure the "IP
10240 service type" value used by the client.
10242 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
10243 request prefix hints in the DHCPv6 solicitation.
10245 * The DHCPv4 server may be configured to send arbitrary options using
10246 a new SendOption= setting.
10248 * The DHCPv4 server may now be configured to emit SIP server list using
10249 the new EmitSIP= and SIP= settings.
10251 * systemd-networkd and networkctl may now renew DHCP leases on demand.
10252 networkctl has a new 'networkctl renew' verb.
10254 * systemd-networkd may now reconfigure links on demand. networkctl
10255 gained two new verbs: "reload" will reload the configuration, and
10256 "reconfigure DEVICE…" will reconfigure one or more devices.
10258 * .network files may now match on SSID and BSSID of a wireless network,
10259 i.e. the access point name and hardware address using the new SSID=
10260 and BSSID= options. networkctl will display the current SSID and
10261 BSSID for wireless links.
10263 .network files may also match on the wireless network type using the
10264 new WLANInterfaceType= option.
10266 * systemd-networkd now includes default configuration that enables
10267 link-local addressing when connected to an ad-hoc wireless network.
10269 * systemd-networkd may configure the Traffic Control queueing
10270 disciplines in the kernel using the new
10271 [TrafficControlQueueingDiscipline] section and Parent=,
10272 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
10273 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
10274 NetworkEmulatorDuplicateRate= settings.
10276 * systemd-tmpfiles gained a new w+ setting to append to files.
10278 * systemd-analyze dump will now report when the memory configuration in
10279 the kernel does not match what systemd has configured (usually,
10280 because some external program has modified the kernel configuration
10283 * systemd-analyze gained a new --base-time= switch instructs the
10284 'calendar' verb to resolve times relative to that timestamp instead
10285 of the present time.
10287 * journalctl --update-catalog now produces deterministic output (making
10288 reproducible image builds easier).
10290 * A new devicetree-overlay setting is now documented in the Boot Loader
10293 * The default value of the WatchdogSec= setting used in systemd
10294 services (the ones bundled with the project itself) may be set at
10295 configuration time using the -Dservice-watchdog= setting. If set to
10296 empty, the watchdogs will be disabled.
10298 * systemd-resolved validates IP addresses in certificates now when GnuTLS
10301 * libcryptsetup >= 2.0.1 is now required.
10303 * A configuration option -Duser-path= may be used to override the $PATH
10304 used by the user service manager. The default is again to use the same
10305 path as the system manager.
10307 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
10308 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
10311 * Service units gained a new sandboxing option ProtectKernelLogs= which
10312 makes sure the program cannot get direct access to the kernel log
10313 buffer anymore, i.e. the syslog() system call (not to be confused
10314 with the API of the same name in libc, which is not affected), the
10315 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
10316 inaccessible to the service. It's recommended to enable this setting
10317 for all services that should not be able to read from or write to the
10318 kernel log buffer, which are probably almost all.
10320 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
10321 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
10322 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
10323 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
10324 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
10325 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
10326 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
10327 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
10328 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
10329 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
10330 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
10331 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
10332 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
10333 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
10334 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
10335 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
10336 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
10337 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
10338 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
10339 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
10340 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
10341 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
10342 Zach Smith, Zbigniew Jędrzejewski-Szmek
10344 – Warsaw, 2019-11-29
10348 * This release enables unprivileged programs (i.e. requiring neither
10349 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
10350 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
10351 kernel for the whole UNIX group range, i.e. all processes. This
10352 change should be reasonably safe, as the kernel support for it was
10353 specifically implemented to allow safe access to ICMP Echo for
10354 processes lacking any privileges. If this is not desirable, it can be
10355 disabled again by setting the parameter to "1 0".
10357 * Previously, filters defined with SystemCallFilter= would have the
10358 effect that any calling of an offending system call would terminate
10359 the calling thread. This behaviour never made much sense, since
10360 killing individual threads of unsuspecting processes is likely to
10361 create more problems than it solves. With this release the default
10362 action changed from killing the thread to killing the whole
10363 process. For this to work correctly both a kernel version (>= 4.14)
10364 and a libseccomp version (>= 2.4.0) supporting this new seccomp
10365 action is required. If an older kernel or libseccomp is used the old
10366 behaviour continues to be used. This change does not affect any
10367 services that have no system call filters defined, or that use
10368 SystemCallErrorNumber= (and thus see EPERM or another error instead
10369 of being killed when calling an offending system call). Note that
10370 systemd documentation always claimed that the whole process is
10371 killed. With this change behaviour is thus adjusted to match the
10374 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
10375 4194304 by default, i.e. the full 22bit range the kernel allows, up
10376 from the old 16-bit range. This should improve security and
10377 robustness, as PID collisions are made less likely (though certainly
10378 still possible). There are rumours this might create compatibility
10379 problems, though at this moment no practical ones are known to
10380 us. Downstream distributions are hence advised to undo this change in
10381 their builds if they are concerned about maximum compatibility, but
10382 for everybody else we recommend leaving the value bumped. Besides
10383 improving security and robustness this should also simplify things as
10384 the maximum number of allowed concurrent tasks was previously bounded
10385 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
10386 only a single knob is left ("kernel.threads-max"). There have been
10387 concerns that usability is affected by this change because larger PID
10388 numbers are harder to type, but we believe the change from 5 digits
10389 to 7 digits doesn't hamper usability.
10391 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
10392 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
10393 hierarchically set default memory protection values for a particular
10394 subtree of the unit hierarchy.
10396 * Memory protection directives can now take a value of zero, allowing
10397 explicit opting out of a default value propagated by an ancestor.
10399 * systemd now defaults to the "unified" cgroup hierarchy setup during
10400 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
10401 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
10402 change reflects the fact that cgroupsv2 support has matured
10403 substantially in both systemd and in the kernel, and is clearly the
10404 way forward. Downstream production distributions might want to
10405 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
10406 their builds as unfortunately the popular container managers have not
10407 caught up with the kernel API changes.
10409 * Man pages are not built by default anymore (html pages were already
10410 disabled by default), to make development builds quicker. When
10411 building systemd for a full installation with documentation, meson
10412 should be called with -Dman=true and/or -Dhtml=true as appropriate.
10413 The default was changed based on the assumption that quick one-off or
10414 repeated development builds are much more common than full optimized
10415 builds for installation, and people need to pass various other
10416 options to when doing "proper" builds anyway, so the gain from making
10417 development builds quicker is bigger than the one time disruption for
10420 Two scripts are created in the *build* directory to generate and
10421 preview man and html pages on demand, e.g.:
10423 build/man/man systemctl
10424 build/man/html systemd.index
10426 * libidn2 is used by default if both libidn2 and libidn are installed.
10427 Please use -Dlibidn=true if libidn is preferred.
10429 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
10430 big-endian machines. Before, bytes were written and read in native
10431 machine order as exposed by the native libc __cpu_mask interface.
10432 Now, little-endian order is always used (CPUs 0–7 are described by
10433 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
10434 This change fixes D-Bus calls that cross endianness boundary.
10436 The presentation format used for CPUAffinity= by "systemctl show" and
10437 "systemd-analyze dump" is changed to present CPU indices instead of
10438 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
10439 shown as CPUAffinity=03000000000000000000000000000… (on
10440 little-endian) or CPUAffinity=00000000000000300000000000000… (on
10441 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
10442 input format. The maximum integer that will be printed in the new
10443 format is 8191 (four digits), while the old format always used a very
10444 long number (with the length varying by architecture), so they can be
10445 unambiguously distinguished.
10447 * /usr/sbin/halt.local is no longer supported. Implementation in
10448 distributions was inconsistent and it seems this functionality was
10451 To replace this functionality, users should:
10452 - either define a new unit and make it a dependency of final.target
10453 (systemctl add-wants final.target my-halt-local.service)
10454 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
10455 and ensure that it accepts "halt", "poweroff", "reboot", and
10456 "kexec" as an argument, see the description in systemd-shutdown(8).
10458 * When a [Match] section in .link or .network file is empty (contains
10459 no match patterns), a warning will be emitted. Please add any "match
10460 all" pattern instead, e.g. OriginalName=* or Name=* in case all
10461 interfaces should really be matched.
10463 * A new setting NUMAPolicy= may be used to set process memory
10464 allocation policy. This setting can be specified in
10465 /etc/systemd/system.conf and hence will set the default policy for
10466 PID1. The default policy can be overridden on a per-service
10467 basis. The related setting NUMAMask= is used to specify NUMA node
10468 mask that should be associated with the selected policy.
10470 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
10471 generates when processes it manages are reaching their memory limits,
10472 and will place their units in a special state, and optionally kill or
10473 stop the whole unit.
10475 * The service manager will now expose bus properties for the IO
10476 resources used by units. This information is also shown in "systemctl
10477 status" now (for services that have IOAccounting=yes set). Moreover,
10478 the IO accounting data is included in the resource log message
10479 generated whenever a unit stops.
10481 * Units may now configure an explicit timeout to wait for when killed
10482 with SIGABRT, for example when a service watchdog is hit. Previously,
10483 the regular TimeoutStopSec= timeout was applied in this case too —
10484 now a separate timeout may be set using TimeoutAbortSec=.
10486 * Services may now send a special WATCHDOG=trigger message with
10487 sd_notify() to trigger an immediate "watchdog missed" event, and thus
10488 trigger service termination. This is useful both for testing watchdog
10489 handling, but also for defining error paths in services, that shall
10490 be handled the same way as watchdog events.
10492 * There are two new per-unit settings IPIngressFilterPath= and
10493 IPEgressFilterPath= which allow configuration of a BPF program
10494 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
10495 to apply to the IP packet ingress/egress path of all processes of a
10496 unit. This is useful to allow running systemd services with BPF
10497 programs set up externally.
10499 * systemctl gained a new "clean" verb for removing the state, cache,
10500 runtime or logs directories of a service while it is terminated. The
10501 new verb may also be used to remove the state maintained on disk for
10502 timer units that have Persistent= configured.
10504 * During the last phase of shutdown systemd will now automatically
10505 increase the log level configured in the "kernel.printk" sysctl so
10506 that any relevant loggable events happening during late shutdown are
10507 made visible. Previously, loggable events happening so late during
10508 shutdown were generally lost if the "kernel.printk" sysctl was set to
10509 high thresholds, as regular logging daemons are terminated at that
10510 time and thus nothing is written to disk.
10512 * If processes terminated during the last phase of shutdown do not exit
10513 quickly systemd will now show their names after a short time, to make
10514 debugging easier. After a longer timeout they are forcibly killed,
10517 * journalctl (and the other tools that display logs) will now highlight
10518 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
10519 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
10520 are now shown in blue color, to separate them visually from regular
10521 logs. References to configuration files are now turned into clickable
10522 links on terminals that support that.
10524 * systemd-journald will now stop logging to /var/log/journal during
10525 shutdown when /var/ is on a separate mount, so that it can be
10526 unmounted safely during shutdown.
10528 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
10530 * systemd-resolved "Cache=" configuration option in resolved.conf has
10531 been extended to also accept the 'no-negative' value. Previously,
10532 only a boolean option was allowed (yes/no), having yes as the
10533 default. If this option is set to 'no-negative', negative answers are
10534 not cached while the old cache heuristics are used positive answers.
10535 The default remains unchanged.
10537 * The predictable naming scheme for network devices now supports
10538 generating predictable names for "netdevsim" devices.
10540 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
10543 Those two changes form a new net.naming_scheme= entry. Distributions
10544 which want to preserve naming stability may want to set the
10545 -Ddefault-net-naming-scheme= configuration option.
10547 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
10548 interfaces natively.
10550 * systemd-networkd's bridge FDB support now allows configuration of a
10551 destination address for each entry (Destination=), as well as the
10552 VXLAN VNI (VNI=), as well as an option to declare what an entry is
10553 associated with (AssociatedWith=).
10555 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
10556 option for configuring the maximum number of DHCP lease requests. It
10557 also learnt a new BlackList= option for deny-listing DHCP servers (a
10558 similar setting has also been added to the IPv6 RA client), as well
10559 as a SendRelease= option for configuring whether to send a DHCP
10560 RELEASE message when terminating.
10562 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
10563 separately in the [DHCPv4] and [DHCPv6] sections.
10565 * systemd-networkd's DHCP support will now optionally create an
10566 implicit host route to the DNS server specified in the DHCP lease, in
10567 addition to the routes listed explicitly in the lease. This should
10568 ensure that in multi-homed systems DNS traffic leaves the systems on
10569 the interface that acquired the DNS server information even if other
10570 routes such as default routes exist. This behaviour may be turned on
10571 with the new RoutesToDNS= option.
10573 * systemd-networkd's VXLAN support gained a new option
10574 GenericProtocolExtension= for enabling VXLAN Generic Protocol
10575 Extension support, as well as IPDoNotFragment= for setting the IP
10576 "Don't fragment" bit on outgoing packets. A similar option has been
10577 added to the GENEVE support.
10579 * In systemd-networkd's [Route] section you may now configure
10580 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
10581 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
10582 propagation. The Type= setting now supports local, broadcast,
10583 anycast, multicast, any, xresolve routes, too.
10585 * systemd-networkd's [Network] section learnt a new option
10586 DefaultRouteOnDevice= for automatically configuring a default route
10587 onto the network device.
10589 * systemd-networkd's bridging support gained two new options ProxyARP=
10590 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
10591 MulticastRouter= for configuring multicast routing behaviour. A new
10592 option MulticastIGMPVersion= may be used to change bridge's multicast
10593 Internet Group Management Protocol (IGMP) version.
10595 * systemd-networkd's FooOverUDP support gained the ability to configure
10596 local and peer IP addresses via Local= and Peer=. A new option
10597 PeerPort= may be used to configure the peer's IP port.
10599 * systemd-networkd's TUN support gained a new setting VnetHeader= for
10600 tweaking Generic Segment Offload support.
10602 * The address family for policy rules may be specified using the new
10603 Family= option in the [RoutingPolicyRule] section.
10605 * networkctl gained a new "delete" command for removing virtual network
10606 devices, as well as a new "--stats" switch for showing device
10609 * networkd.conf gained a new setting SpeedMeter= and
10610 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
10611 measured speed may be shown by 'networkctl status'.
10613 * "networkctl status" now displays MTU and queue lengths, and more
10614 detailed information about VXLAN and bridge devices.
10616 * systemd-networkd's .network and .link files gained a new Property=
10617 setting in the [Match] section, to match against devices with
10618 specific udev properties.
10620 * systemd-networkd's tunnel support gained a new option
10621 AssignToLoopback= for selecting whether to use the loopback device
10622 "lo" as underlying device.
10624 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
10625 been renamed to LinkLayerAddress=, and it now allows configuration of
10628 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
10629 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
10630 IPv6 configuration (static or DHCPv6) was found for a given
10631 interface. It will not touch the sysctl otherwise.
10633 * The order of entries is $PATH used by the user manager instance was
10634 changed to put bin/ entries before the corresponding sbin/ entries.
10635 It is recommended to not rely on this order, and only ever have one
10636 binary with a given name in the system paths under /usr.
10638 * A new tool systemd-network-generator has been added that may generate
10639 .network, .netdev and .link files from IP configuration specified on
10640 the kernel command line in the format used by Dracut.
10642 * The CriticalConnection= setting in .network files is now deprecated,
10643 and replaced by a new KeepConfiguration= setting which allows more
10644 detailed configuration of the IP configuration to keep in place.
10646 * systemd-analyze gained a few new verbs:
10648 - "systemd-analyze timestamp" parses and converts timestamps. This is
10649 similar to the existing "systemd-analyze calendar" command which
10650 does the same for recurring calendar events.
10652 - "systemd-analyze timespan" parses and converts timespans (i.e.
10653 durations as opposed to points in time).
10655 - "systemd-analyze condition" will parse and test ConditionXYZ=
10658 - "systemd-analyze exit-status" will parse and convert exit status
10659 codes to their names and back.
10661 - "systemd-analyze unit-files" will print a list of all unit
10662 file paths and unit aliases.
10664 * SuccessExitStatus=, RestartPreventExitStatus=, and
10665 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
10666 is equivalent to "65"). Those exit status name mappings may be
10667 displayed with the systemd-analyze exit-status verb describe above.
10669 * systemd-logind now exposes a per-session SetBrightness() bus call,
10670 which may be used to securely change the brightness of a kernel
10671 brightness device, if it belongs to the session's seat. By using this
10672 call unprivileged clients can make changes to "backlight" and "leds"
10673 devices securely with strict requirements on session membership.
10674 Desktop environments may use this to generically make brightness
10675 changes to such devices without shipping private SUID binaries or
10676 udev rules for that purpose.
10678 * "udevadm info" gained a --wait-for-initialization switch to wait for
10679 a device to be initialized.
10681 * systemd-hibernate-resume-generator will now look for resumeflags= on
10682 the kernel command line, which is similar to rootflags= and may be
10683 used to configure device timeout for the hibernation device.
10685 * sd-event learnt a new API call sd_event_source_disable_unref() for
10686 disabling and unref'ing an event source in a single function. A
10687 related call sd_event_source_disable_unrefp() has been added for use
10688 with gcc's cleanup extension.
10690 * The sd-id128.h public API gained a new definition
10691 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
10694 * "busctl introspect" gained a new switch --xml-interface for dumping
10695 XML introspection data unmodified.
10697 * PID 1 may now show the unit name instead of the unit description
10698 string in its status output during boot. This may be configured in
10699 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
10700 kernel command line option systemd.status_unit_format=.
10702 * PID 1 now understands a new option KExecWatchdogSec= in
10703 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
10704 Previously watchdog functionality was only available for regular
10705 reboots. The new setting defaults to off, because we don't know in
10706 the general case if the watchdog will be reset after kexec (some
10707 drivers do reset it, but not all), and the new userspace might not be
10708 configured to handle the watchdog.
10710 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
10711 RebootWatchdogSec= to more clearly communicate what it is about. The
10712 old name is still accepted for compatibility.
10714 * The systemd.debug_shell kernel command line option now optionally
10715 takes a tty name to spawn the debug shell on, which allows a
10716 different tty to be selected than the built-in default.
10718 * Service units gained a new ExecCondition= setting which will run
10719 before ExecStartPre= and either continue execution of the unit (for
10720 clean exit codes), stop execution without marking the unit failed
10721 (for exit codes 1 through 254), or stop execution and fail the unit
10722 (for exit code 255 or abnormal termination).
10724 * A new service systemd-pstore.service has been added that pulls data
10725 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
10728 * timedatectl gained new verbs for configuring per-interface NTP
10729 service configuration for systemd-timesyncd.
10731 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
10732 2019. (You can set non-UTF-8 locales though, if you know their name.)
10734 * If variable assignments in sysctl.d/ files are prefixed with "-" any
10735 failures to apply them are now ignored.
10737 * systemd-random-seed.service now optionally credits entropy when
10738 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
10739 true for the service to enable this behaviour, but please consult the
10740 documentation first, since this comes with a couple of caveats.
10742 * systemd-random-seed.service is now a synchronization point for full
10743 initialization of the kernel's entropy pool. Services that require
10744 /dev/urandom to be correctly initialized should be ordered after this
10747 * The systemd-boot boot loader has been updated to optionally maintain
10748 a random seed file in the EFI System Partition (ESP). During the boot
10749 phase, this random seed is read and updated with a new seed
10750 cryptographically derived from it. Another derived seed is passed to
10751 the OS. The latter seed is then credited to the kernel's entropy pool
10752 very early during userspace initialization (from PID 1). This allows
10753 systems to boot up with a fully initialized kernel entropy pool from
10754 earliest boot on, and thus entirely removes all entropy pool
10755 initialization delays from systems using systemd-boot. Special care
10756 is taken to ensure different seeds are derived on system images
10757 replicated to multiple systems. "bootctl status" will show whether
10758 a seed was received from the boot loader.
10760 * bootctl gained two new verbs:
10762 - "bootctl random-seed" will generate the file in ESP and an EFI
10763 variable to allow a random seed to be passed to the OS as described
10766 - "bootctl is-installed" checks whether systemd-boot is currently
10769 * bootctl will warn if it detects that boot entries are misconfigured
10770 (for example if the kernel image was removed without purging the
10773 * A new document has been added describing systemd's use and support
10774 for the kernel's entropy pool subsystem:
10776 https://systemd.io/RANDOM_SEEDS
10778 * When the system is hibernated the swap device to write the
10779 hibernation image to is now automatically picked from all available
10780 swap devices, preferring the swap device with the highest configured
10781 priority over all others, and picking the device with the most free
10782 space if there are multiple devices with the highest priority.
10784 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
10785 option that permits selecting the timeout how long to wait for a
10786 device with an encryption key before asking for the password.
10788 * IOWeight= has learnt to properly set the IO weight when using the
10789 BFQ scheduler officially found in kernels 5.0+.
10791 * A new mailing list has been created for reporting of security issues:
10792 systemd-security@redhat.com. For mode details, see
10793 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
10795 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
10796 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
10797 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
10798 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
10799 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
10800 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
10801 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
10802 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
10803 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
10804 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
10805 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
10806 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
10807 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
10808 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
10809 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
10810 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
10811 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
10812 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
10813 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
10814 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
10815 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
10816 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
10817 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
10818 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
10819 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
10820 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
10821 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
10822 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
10823 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
10824 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
10826 – Camerino, 2019-09-03
10830 * In .link files, MACAddressPolicy=persistent (the default) is changed
10831 to cover more devices. For devices like bridges, tun, tap, bond, and
10832 similar interfaces that do not have other identifying information,
10833 the interface name is used as the basis for persistent seed for MAC
10834 and IPv4LL addresses. The way that devices that were handled
10835 previously is not changed, and this change is about covering more
10836 devices then previously by the "persistent" policy.
10838 MACAddressPolicy=random may be used to force randomized MACs and
10839 IPv4LL addresses for a device if desired.
10841 Hint: the log output from udev (at debug level) was enhanced to
10842 clarify what policy is followed and which attributes are used.
10843 'SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>'
10844 may be used to view this.
10846 Hint: if a bridge interface is created without any slaves, and gains
10847 a slave later, then now the bridge does not inherit slave's MAC.
10848 To inherit slave's MAC, for example, create the following file:
10850 # /etc/systemd/network/98-bridge-inherit-mac.link
10855 MACAddressPolicy=none
10858 * The .device units generated by systemd-fstab-generator and other
10859 generators do not automatically pull in the corresponding .mount unit
10860 as a Wants= dependency. This means that simply plugging in the device
10861 will not cause the mount unit to be started automatically. But please
10862 note that the mount unit may be started for other reasons, in
10863 particular if it is part of local-fs.target, and any unit which
10864 (transitively) depends on local-fs.target is started.
10866 * networkctl list/status/lldp now accept globbing wildcards for network
10867 interface names to match against all existing interfaces.
10869 * The $PIDFILE environment variable is set to point the absolute path
10870 configured with PIDFile= for processes of that service.
10872 * The fallback DNS server list was augmented with Cloudflare public DNS
10873 servers. Use '-Ddns-servers=' to set a different fallback.
10875 * A new special target usb-gadget.target will be started automatically
10876 when a USB Device Controller is detected (which means that the system
10877 is a USB peripheral).
10879 * A new unit setting CPUQuotaPeriodSec= assigns the time period
10880 relatively to which the CPU time quota specified by CPUQuota= is
10883 * A new unit setting ProtectHostname= may be used to prevent services
10884 from modifying hostname information (even if they otherwise would
10885 have privileges to do so).
10887 * A new unit setting NetworkNamespacePath= may be used to specify a
10888 namespace for service or socket units through a path referring to a
10889 Linux network namespace pseudo-file.
10891 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
10892 have an effect on .socket units: when used the listening socket is
10893 created within the configured network namespace instead of the host
10896 * ExecStart= command lines in unit files may now be prefixed with ':'
10897 in which case environment variable substitution is
10898 disabled. (Supported for the other ExecXYZ= settings, too.)
10900 * .timer units gained two new boolean settings OnClockChange= and
10901 OnTimezoneChange= which may be used to also trigger a unit when the
10902 system clock is changed or the local timezone is
10903 modified. systemd-run has been updated to make these options easily
10904 accessible from the command line for transient timers.
10906 * Two new conditions for units have been added: ConditionMemory= may be
10907 used to conditionalize a unit based on installed system
10908 RAM. ConditionCPUs= may be used to conditionalize a unit based on
10909 installed CPU cores.
10911 * The @default system call filter group understood by SystemCallFilter=
10912 has been updated to include the new rseq() system call introduced in
10915 * A new time-set.target has been added that indicates that the system
10916 time has been set from a local source (possibly imprecise). The
10917 existing time-sync.target is stronger and indicates that the time has
10918 been synchronized with a precise external source. Services where
10919 approximate time is sufficient should use the new target.
10921 * "systemctl start" (and related commands) learnt a new
10922 --show-transaction option. If specified brief information about all
10923 jobs queued because of the requested operation is shown.
10925 * systemd-networkd recognizes a new operation state 'enslaved', used
10926 (instead of 'degraded' or 'carrier') for interfaces which form a
10927 bridge, bond, or similar, and an new 'degraded-carrier' operational
10928 state used for the bond or bridge master interface when one of the
10929 enslaved devices is not operational.
10931 * .network files learnt the new IgnoreCarrierLoss= option for leaving
10932 networks configured even if the carrier is lost.
10934 * The RequiredForOnline= setting in .network files may now specify a
10935 minimum operational state required for the interface to be considered
10936 "online" by systemd-networkd-wait-online. Related to this
10937 systemd-networkd-wait-online gained a new option --operational-state=
10938 to configure the same, and its --interface= option was updated to
10939 optionally also take an operational state specific for an interface.
10941 * systemd-networkd-wait-online gained a new setting --any for waiting
10942 for only one of the requested interfaces instead of all of them.
10944 * systemd-networkd now implements L2TP tunnels.
10946 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
10947 may be used to cause autonomous and onlink prefixes received in IPv6
10948 Router Advertisements to be ignored.
10950 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
10951 file settings may be used to tweak bridge behaviour.
10953 * The new TripleSampling= option in .network files may be used to
10954 configure CAN triple sampling.
10956 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
10957 used to point to private or preshared key for a WireGuard interface.
10959 * /etc/crypttab now supports the same-cpu-crypt and
10960 submit-from-crypt-cpus options to tweak encryption work scheduling
10963 * systemd-tmpfiles will now take a BSD file lock before operating on a
10964 contents of directory. This may be used to temporarily exclude
10965 directories from aging by taking the same lock (useful for example
10966 when extracting a tarball into /tmp or /var/tmp as a privileged user,
10967 which might create files with really old timestamps, which
10968 nevertheless should not be deleted). For further details, see:
10970 https://systemd.io/TEMPORARY_DIRECTORIES
10972 * systemd-tmpfiles' h line type gained support for the
10973 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
10974 controlling project quota inheritance.
10976 * sd-boot and bootctl now implement support for an Extended Boot Loader
10977 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
10978 addition to the ESP partition mounted to /efi or /boot/efi.
10979 Configuration file fragments, kernels, initrds and other EFI images
10980 to boot will be loaded from both the ESP and XBOOTLDR partitions.
10981 The XBOOTLDR partition was previously described by the Boot Loader
10982 Specification, but implementation was missing in sd-boot. Support for
10983 this concept allows using the sd-boot boot loader in more
10984 conservative scenarios where the boot loader itself is placed in the
10985 ESP but the kernels to boot (and their metadata) in a separate
10988 * A system may now be booted with systemd.volatile=overlay on the
10989 kernel command line, which causes the root file system to be set up
10990 an overlayfs mount combining the root-only root directory with a
10991 writable tmpfs. In this setup, the underlying root device is not
10992 modified, and any changes are lost at reboot.
10994 * Similar, systemd-nspawn can now boot containers with a volatile
10995 overlayfs root with the new --volatile=overlay switch.
10997 * systemd-nspawn can now consume OCI runtime bundles using a new
10998 --oci-bundle= option. This implementation is fully usable, with most
10999 features in the specification implemented, but since this a lot of
11000 new code and functionality, this feature should most likely not
11001 be used in production yet.
11003 * systemd-nspawn now supports various options described by the OCI
11004 runtime specification on the command-line and in .nspawn files:
11005 --inaccessible=/Inaccessible= may be used to mask parts of the file
11006 system tree, --console=/--pipe may be used to configure how standard
11007 input, output, and error are set up.
11009 * busctl learned the 'emit' verb to generate D-Bus signals.
11011 * systemd-analyze cat-config may be used to gather and display
11012 configuration spread over multiple files, for example system and user
11013 presets, tmpfiles.d, sysusers.d, udev rules, etc.
11015 * systemd-analyze calendar now takes an optional new parameter
11016 --iterations= which may be used to show a maximum number of iterations
11017 the specified expression will elapse next.
11019 * The sd-bus C API gained support for naming method parameters in the
11020 introspection data.
11022 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
11023 the reboot() system call expects.
11025 * journalctl learnt a new --cursor-file= option that points to a file
11026 from which a cursor should be loaded in the beginning and to which
11027 the updated cursor should be stored at the end.
11029 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
11030 detected by systemd-detect-virt (and may also be used in
11031 ConditionVirtualization=).
11033 * The behaviour of systemd-logind may now be modified with environment
11034 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
11035 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
11036 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
11037 skip the relevant operation completely (when set to false), or to
11038 create a flag file in /run/systemd (when set to true), instead of
11039 actually commencing the real operation when requested. The presence
11040 of /run/systemd/reboot-to-firmware-setup,
11041 /run/systemd/reboot-to-boot-loader-menu, and
11042 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
11043 boot loader implementations to replace some steps logind performs
11044 during reboot with their own operations.
11046 * systemctl can be used to request a reboot into the boot loader menu
11047 or a specific boot loader entry with the new --boot-load-menu= and
11048 --boot-loader-entry= options to a reboot command. (This requires a
11049 boot loader that supports this, for example sd-boot.)
11051 * kernel-install will no longer unconditionally create the output
11052 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
11053 snippets, but will do only if the machine-specific parent directory
11054 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
11055 to create this parent directory during sd-boot installation.
11057 This makes it easier to use kernel-install with plugins which support
11058 a different layout of the bootloader partitions (for example grub2).
11060 * During package installation (with 'ninja install'), we would create
11061 symlinks for getty@tty1.service, systemd-networkd.service,
11062 systemd-networkd.socket, systemd-resolved.service,
11063 remote-cryptsetup.target, remote-fs.target,
11064 systemd-networkd-wait-online.service, and systemd-timesyncd.service
11065 in /etc, as if 'systemctl enable' was called for those units, to make
11066 the system usable immediately after installation. Now this is not
11067 done anymore, and instead calling 'systemctl preset-all' is
11068 recommended after the first installation of systemd.
11070 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
11071 is built on seccomp. When turned on creation of SUID/SGID files is
11074 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
11075 implied if DynamicUser= is turned on for a service. This hardens
11076 these services, so that they neither can benefit from nor create
11077 SUID/SGID executables. This is a minor compatibility breakage, given
11078 that when DynamicUser= was first introduced SUID/SGID behaviour was
11079 unaffected. However, the security benefit of these two options is
11080 substantial, and the setting is still relatively new, hence we opted
11081 to make it mandatory for services with dynamic users.
11083 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
11084 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
11085 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
11086 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
11087 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
11088 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
11089 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
11090 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
11091 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
11092 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
11093 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
11094 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
11095 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
11096 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
11097 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
11098 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
11099 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
11100 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
11102 — Warsaw, 2019-04-11
11106 * The default locale can now be configured at compile time. Otherwise,
11107 a suitable default will be selected automatically (one of C.UTF-8,
11108 en_US.UTF-8, and C).
11110 * The version string shown by systemd and other tools now includes the
11111 git commit hash when built from git. An override may be specified
11112 during compilation, which is intended to be used by distributions to
11113 include the package release information.
11115 * systemd-cat can now filter standard input and standard error streams
11116 for different syslog priorities using the new --stderr-priority=
11119 * systemd-journald and systemd-journal-remote reject entries which
11120 contain too many fields (CVE-2018-16865) and set limits on the
11121 process' command line length (CVE-2018-16864).
11123 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
11126 * A new network device NamePolicy "keep" is implemented for link files,
11127 and used by default in 99-default.link (the fallback configuration
11128 provided by systemd). With this policy, if the network device name
11129 was already set by userspace, the device will not be renamed again.
11130 This matches the naming scheme that was implemented before
11131 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
11132 is also enabled by default, even if not specified. Effectively, this
11133 means that if naming-scheme >= 240 is specified, network devices will
11134 be renamed according to the configuration, even if they have been
11135 renamed already, if "keep" is not specified as the naming policy in
11136 the .link file. The 99-default.link file provided by systemd includes
11137 "keep" for backwards compatibility, but it is recommended for user
11138 installed .link files to *not* include it.
11140 The "kernel" policy, which keeps kernel names declared to be
11141 "persistent", now works again as documented.
11143 * kernel-install script now optionally takes the paths to one or more
11144 initrd files, and passes them to all plugins.
11146 * The mincore() system call has been dropped from the @system-service
11147 system call filter group, as it is pretty exotic and may potentially
11148 used for side-channel attacks.
11150 * -fPIE is dropped from compiler and linker options. Please specify
11151 -Db_pie=true option to meson to build position-independent
11152 executables. Note that the meson option is supported since meson-0.49.
11154 * The fs.protected_regular and fs.protected_fifos sysctls, which were
11155 added in Linux 4.19 to make some data spoofing attacks harder, are
11156 now enabled by default. While this will hopefully improve the
11157 security of most installations, it is technically a backwards
11158 incompatible change; to disable these sysctls again, place the
11159 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
11161 fs.protected_regular = 0
11162 fs.protected_fifos = 0
11164 Note that the similar hardlink and symlink protection has been
11165 enabled since v199, and may be disabled likewise.
11167 * The files read from the EnvironmentFile= setting in unit files now
11168 parse backslashes inside quotes literally, matching the behaviour of
11171 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
11172 now automatically become NOPs when run in a chroot() environment.
11174 * The tmpfiles.d/ "C" line type will now copy directory trees not only
11175 when the destination is so far missing, but also if it already exists
11176 as a directory and is empty. This is useful to cater for systems
11177 where directory trees are put together from multiple separate mount
11178 points but otherwise empty.
11180 * A new function sd_bus_close_unref() (and the associated
11181 sd_bus_close_unrefp()) has been added to libsystemd, that combines
11182 sd_bus_close() and sd_bus_unref() in one.
11184 * udevadm control learnt a new option for --ping for testing whether a
11185 systemd-udevd instance is running and reacting.
11187 * udevadm trigger learnt a new option for --wait-daemon for waiting
11188 systemd-udevd daemon to be initialized.
11190 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
11191 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
11192 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
11193 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
11194 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
11195 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
11196 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
11197 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
11198 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
11199 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
11200 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
11201 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
11202 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
11203 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
11204 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
11205 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
11206 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
11208 — Berlin, 2019-02-14
11212 * NoNewPrivileges=yes has been set for all long-running services
11213 implemented by systemd. Previously, this was problematic due to
11214 SELinux (as this would also prohibit the transition from PID1's label
11215 to the service's label). This restriction has since been lifted, but
11216 an SELinux policy update is required.
11217 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
11219 * DynamicUser=yes is dropped from systemd-networkd.service,
11220 systemd-resolved.service and systemd-timesyncd.service, which was
11221 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
11222 and since v236 for systemd-timesyncd.service. The users and groups
11223 systemd-network, systemd-resolve and systemd-timesync are created
11224 by systemd-sysusers again. Distributors or system administrators
11225 may need to create these users and groups if they not exist (or need
11226 to re-enable DynamicUser= for those units) while upgrading systemd.
11227 Also, the clock file for systemd-timesyncd may need to move from
11228 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
11230 * When unit files are loaded from disk, previously systemd would
11231 sometimes (depending on the unit loading order) load units from the
11232 target path of symlinks in .wants/ or .requires/ directories of other
11233 units. This meant that unit could be loaded from different paths
11234 depending on whether the unit was requested explicitly or as a
11235 dependency of another unit, not honouring the priority of directories
11236 in search path. It also meant that it was possible to successfully
11237 load and start units which are not found in the unit search path, as
11238 long as they were requested as a dependency and linked to from
11239 .wants/ or .requires/. The target paths of those symlinks are not
11240 used for loading units anymore and the unit file must be found in
11243 * A new service type has been added: Type=exec. It's very similar to
11244 Type=simple but ensures the service manager will wait for both fork()
11245 and execve() of the main service binary to complete before proceeding
11246 with follow-up units. This is primarily useful so that the manager
11247 propagates any errors in the preparation phase of service execution
11248 back to the job that requested the unit to be started. For example,
11249 consider a service that has ExecStart= set to a file system binary
11250 that doesn't exist. With Type=simple starting the unit would be
11251 considered instantly successful, as only fork() has to complete
11252 successfully and the manager does not wait for execve(), and hence
11253 its failure is seen "too late". With the new Type=exec service type
11254 starting the unit will fail, as the manager will wait for the
11255 execve() and notice its failure, which is then propagated back to the
11258 NOTE: with the next release 241 of systemd we intend to change the
11259 systemd-run tool to default to Type=exec for transient services
11260 started by it. This should be mostly safe, but in specific corner
11261 cases might result in problems, as the systemd-run tool will then
11262 block on NSS calls (such as user name look-ups due to User=) done
11263 between the fork() and execve(), which under specific circumstances
11264 might cause problems. It is recommended to specify "-p Type=simple"
11265 explicitly in the few cases where this applies. For regular,
11266 non-transient services (i.e. those defined with unit files on disk)
11267 we will continue to default to Type=simple.
11269 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
11270 userspace processes is set to 1024 (soft) and 4096
11271 (hard). Previously, systemd passed this on unmodified to all
11272 processes it forked off. With this systemd release the hard limit
11273 systemd passes on is increased to 512K, overriding the kernel's
11274 defaults and substantially increasing the number of simultaneous file
11275 descriptors unprivileged userspace processes can allocate. Note that
11276 the soft limit remains at 1024 for compatibility reasons: the
11277 traditional UNIX select() call cannot deal with file descriptors >=
11278 1024 and increasing the soft limit globally might thus result in
11279 programs unexpectedly allocating a high file descriptor and thus
11280 failing abnormally when attempting to use it with select() (of
11281 course, programs shouldn't use select() anymore, and prefer
11282 poll()/epoll, but the call unfortunately remains undeservedly popular
11283 at this time). This change reflects the fact that file descriptor
11284 handling in the Linux kernel has been optimized in more recent
11285 kernels and allocating large numbers of them should be much cheaper
11286 both in memory and in performance than it used to be. Programs that
11287 want to take benefit of the increased limit have to "opt-in" into
11288 high file descriptors explicitly by raising their soft limit. Of
11289 course, when they do that they must acknowledge that they cannot use
11290 select() anymore (and neither can any shared library they use — or
11291 any shared library used by any shared library they use and so on).
11292 Which default hard limit is most appropriate is of course hard to
11293 decide. However, given reports that ~300K file descriptors are used
11294 in real-life applications we believe 512K is sufficiently high as new
11295 default for now. Note that there are also reports that using very
11296 high hard limits (e.g. 1G) is problematic: some software allocates
11297 large arrays with one element for each potential file descriptor
11298 (Java, …) — a high hard limit thus triggers excessively large memory
11299 allocations in these applications. Hopefully, the new default of 512K
11300 is a good middle ground: higher than what real-life applications
11301 currently need, and low enough for avoid triggering excessively large
11302 allocations in problematic software. (And yes, somebody should fix
11305 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
11306 to the highest possible values, as separate accounting of file
11307 descriptors is no longer necessary, as memcg tracks them correctly as
11308 part of the memory accounting anyway. Thus, from the four limits on
11309 file descriptors currently enforced (fs.file-max, fs.nr_open,
11310 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
11311 and keep only the latter two. A set of build-time options
11312 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
11313 has been added to revert this change in behaviour, which might be
11314 an option for systems that turn off memcg in the kernel.
11316 * When no /etc/locale.conf file exists (and hence no locale settings
11317 are in place), systemd will now use the "C.UTF-8" locale by default,
11318 and set LANG= to it. This locale is supported by various
11319 distributions including Fedora, with clear indications that upstream
11320 glibc is going to make it available too. This locale enables UTF-8
11321 mode by default, which appears appropriate for 2018.
11323 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
11324 default. This effectively switches the RFC3704 Reverse Path filtering
11325 from Strict mode to Loose mode. This is more appropriate for hosts
11326 that have multiple links with routes to the same networks (e.g.
11327 a client with a Wi-Fi and Ethernet both connected to the internet).
11329 Consult the kernel documentation for details on this sysctl:
11330 https://docs.kernel.org/networking/ip-sysctl.html
11332 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
11335 * CPUAccounting=yes no longer enables the CPU controller when using
11336 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
11337 statistics are now provided independently from the CPU controller.
11339 * Support for disabling a particular cgroup controller within a sub-tree
11340 has been added through the DisableControllers= directive.
11342 * cgroup_no_v1=all on the kernel command line now also implies
11343 using the unified cgroup hierarchy, unless one explicitly passes
11344 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
11346 * The new "MemoryMin=" unit file property may now be used to set the
11347 memory usage protection limit of processes invoked by the unit. This
11348 controls the cgroup v2 memory.min attribute. Similarly, the new
11349 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
11350 cgroup v2 io.latency cgroup property for configuring per-service I/O
11353 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
11354 to the cgroup v1 "devices" cgroup controller.
11356 * systemd-escape now is able to combine --unescape with --template. It
11357 also learnt a new option --instance for extracting and unescaping the
11358 instance part of a unit name.
11360 * sd-bus now provides the sd_bus_message_readv() which is similar to
11361 sd_bus_message_read() but takes a va_list object. The pair
11362 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
11363 has been added for configuring the default method call timeout to
11364 use. sd_bus_error_move() may be used to efficiently move the contents
11365 from one sd_bus_error structure to another, invalidating the
11366 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
11367 be used to control whether a bus connection object is automatically
11368 flushed when an sd-event loop is exited.
11370 * When processing classic BSD syslog log messages, journald will now
11371 save the original time-stamp string supplied in the new
11372 SYSLOG_TIMESTAMP= journal field. This permits consumers to
11373 reconstruct the original BSD syslog message more correctly.
11375 * StandardOutput=/StandardError= in service files gained support for
11376 new "append:…" parameters, for connecting STDOUT/STDERR of a service
11377 to a file, and appending to it.
11379 * The signal to use as last step of killing of unit processes is now
11380 configurable. Previously it was hard-coded to SIGKILL, which may now
11381 be overridden with the new KillSignal= setting. Note that this is the
11382 signal used when regular termination (i.e. SIGTERM) does not suffice.
11383 Similarly, the signal used when aborting a program in case of a
11384 watchdog timeout may now be configured too (WatchdogSignal=).
11386 * The XDG_SESSION_DESKTOP environment variable may now be configured in
11387 the pam_systemd argument line, using the new desktop= switch. This is
11388 useful to initialize it properly from a display manager without
11389 having to touch C code.
11391 * Most configuration options that previously accepted percentage values
11392 now also accept permille values with the '‰' suffix (instead of '%').
11394 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
11397 * systemd-resolved's configuration file resolved.conf gained a new
11398 option ReadEtcHosts= which may be used to turn off processing and
11399 honoring /etc/hosts entries.
11401 * The "--wait" switch may now be passed to "systemctl
11402 is-system-running", in which case the tool will synchronously wait
11403 until the system finished start-up.
11405 * hostnamed gained a new bus call to determine the DMI product UUID.
11407 * On x86-64 systemd will now prefer using the RDRAND processor
11408 instruction over /dev/urandom whenever it requires randomness that
11409 neither has to be crypto-grade nor should be reproducible. This
11410 should substantially reduce the amount of entropy systemd requests
11411 from the kernel during initialization on such systems, though not
11412 reduce it to zero. (Why not zero? systemd still needs to allocate
11413 UUIDs and such uniquely, which require high-quality randomness.)
11415 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
11416 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
11417 for forcing the "Other Information" bit in IPv6 RA messages. The
11418 bonding logic gained four new options AdActorSystemPriority=,
11419 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
11420 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
11421 shuffling of flows. The tunnel logic gained a new
11422 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
11423 Deployment. The policy rule logic gained four new options IPProtocol=,
11424 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
11425 support for the MulticastToUnicast= option. networkd also gained
11426 support for configuring static IPv4 ARP or IPv6 neighbor entries.
11428 * .preset files (as read by 'systemctl preset') may now be used to
11429 instantiate services.
11431 * /etc/crypttab now understands the sector-size= option to configure
11432 the sector size for an encrypted partition.
11434 * Key material for encrypted disks may now be placed on a formatted
11435 medium, and referenced from /etc/crypttab by the UUID of the file
11436 system, followed by "=" suffixed by the path to the key file.
11438 * The "collect" udev component has been removed without replacement, as
11439 it is neither used nor maintained.
11441 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
11442 LogsDirectory=, ConfigurationDirectory= settings are used in a
11443 service the executed processes will now receive a set of environment
11444 variables containing the full paths of these directories.
11445 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
11446 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
11447 are used. Note that these options may be used multiple times per
11448 service in which case the resulting paths will be concatenated and
11449 separated by colons.
11451 * Predictable interface naming has been extended to cover InfiniBand
11452 NICs. They will be exposed with an "ib" prefix.
11454 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
11455 which case the respective line failing is ignored.
11457 * .link files may now be used to configure the equivalent to the
11458 "ethtool advertise" commands.
11460 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
11461 alternative to libudev.h. Previously, the latter was just an internal
11462 wrapper around the former, but now these two APIs are exposed
11465 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
11466 which calculates an app-specific boot ID similar to how
11467 sd_id128_get_machine_app_specific() generates an app-specific machine
11470 * A new tool systemd-id128 has been added that can be used to determine
11471 and generate various 128-bit IDs.
11473 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
11476 * systemd-hibernate-resume-generator will now honor the "noresume"
11477 kernel command line option, in which case it will bypass resuming
11478 from any hibernated image.
11480 * The systemd-sleep.conf configuration file gained new options
11481 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
11482 AllowHybridSleep= for prohibiting specific sleep modes even if the
11483 kernel exports them.
11485 * portablectl is now officially supported and has thus moved to
11488 * bootctl learnt the two new commands "set-default" and "set-oneshot"
11489 for setting the default boot loader item to boot to (either
11490 persistently or only for the next boot). This is currently only
11491 compatible with sd-boot, but may be implemented on other boot loaders
11492 too, that follow the boot loader interface. The updated interface is
11493 now documented here:
11495 https://systemd.io/BOOT_LOADER_INTERFACE
11497 * A new kernel command line option systemd.early_core_pattern= is now
11498 understood which may be used to influence the core_pattern PID 1
11499 installs during early boot.
11501 * busctl learnt two new options -j and --json= for outputting method
11502 call replies, properties and monitoring output in JSON.
11504 * journalctl's JSON output now supports simple ANSI coloring as well as
11505 a new "json-seq" mode for generating RFC7464 output.
11507 * Unit files now support the %g/%G specifiers that resolve to the UNIX
11508 group/GID of the service manager runs as, similar to the existing
11509 %u/%U specifiers that resolve to the UNIX user/UID.
11511 * systemd-logind learnt a new global configuration option
11512 UserStopDelaySec= that may be set in logind.conf. It specifies how
11513 long the systemd --user instance shall remain started after a user
11514 logs out. This is useful to speed up repetitive re-connections of the
11515 same user, as it means the user's service manager doesn't have to be
11516 stopped/restarted on each iteration, but can be reused between
11517 subsequent options. This setting defaults to 10s. systemd-logind also
11518 exports two new properties on its Manager D-Bus objects indicating
11519 whether the system's lid is currently closed, and whether the system
11522 * systemd gained support for a generic boot counting logic, which
11523 generically permits automatic reverting to older boot loader entries
11524 if newer updated ones don't work. The boot loader side is implemented
11525 in sd-boot, but is kept open for other boot loaders too. For details
11528 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
11530 * The SuccessAction=/FailureAction= unit file settings now learnt two
11531 new parameters: "exit" and "exit-force", which result in immediate
11532 exiting of the service manager, and are only useful in systemd --user
11533 and container environments.
11535 * Unit files gained support for a pair of options
11536 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
11537 exit status to use as service manager exit status when
11538 SuccessAction=/FailureAction= is set to exit or exit-force.
11540 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
11541 options may now be used to configure the log rate limiting applied by
11542 journald per-service.
11544 * systemd-analyze gained a new verb "timespan" for parsing and
11545 normalizing time span values (i.e. strings like "5min 7s 8us").
11547 * systemd-analyze also gained a new verb "security" for analyzing the
11548 security and sand-boxing settings of services in order to determine an
11549 "exposure level" for them, indicating whether a service would benefit
11550 from more sand-boxing options turned on for them.
11552 * "systemd-analyze syscall-filter" will now also show system calls
11553 supported by the local kernel but not included in any of the defined
11556 * .nspawn files now understand the Ephemeral= setting, matching the
11557 --ephemeral command line switch.
11559 * sd-event gained the new APIs sd_event_source_get_floating() and
11560 sd_event_source_set_floating() for controlling whether a specific
11561 event source is "floating", i.e. destroyed along with the even loop
11564 * Unit objects on D-Bus gained a new "Refs" property that lists all
11565 clients that currently have a reference on the unit (to ensure it is
11568 * The JoinControllers= option in system.conf is no longer supported, as
11569 it didn't work correctly, is hard to support properly, is legacy (as
11570 the concept only exists on cgroup v1) and apparently wasn't used.
11572 * Journal messages that are generated whenever a unit enters the failed
11573 state are now tagged with a unique MESSAGE_ID. Similarly, messages
11574 generated whenever a service process exits are now made recognizable,
11575 too. A tagged message is also emitted whenever a unit enters the
11576 "dead" state on success.
11578 * systemd-run gained a new switch --working-directory= for configuring
11579 the working directory of the service to start. A shortcut -d is
11580 equivalent, setting the working directory of the service to the
11581 current working directory of the invoking program. The new --shell
11582 (or just -S) option has been added for invoking the $SHELL of the
11583 caller as a service, and implies --pty --same-dir --wait --collect
11584 --service-type=exec. Or in other words, "systemd-run -S" is now the
11585 quickest way to quickly get an interactive in a fully clean and
11586 well-defined system service context.
11588 * machinectl gained a new verb "import-fs" for importing an OS tree
11589 from a directory. Moreover, when a directory or tarball is imported
11590 and single top-level directory found with the OS itself below the OS
11591 tree is automatically mangled and moved one level up.
11593 * systemd-importd will no longer set up an implicit btrfs loop-back
11594 file system on /var/lib/machines. If one is already set up, it will
11595 continue to be used.
11597 * A new generator "systemd-run-generator" has been added. It will
11598 synthesize a unit from one or more program command lines included in
11599 the kernel command line. This is very useful in container managers
11602 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
11604 This will run "systemd-nspawn" on an image, invoke the specified
11605 command line and immediately shut down the container again, returning
11606 the command line's exit code.
11608 * The block device locking logic is now documented:
11610 https://systemd.io/BLOCK_DEVICE_LOCKING
11612 * loginctl and machinectl now optionally output the various tables in
11613 JSON using the --output= switch. It is our intention to add similar
11614 support to systemctl and all other commands.
11616 * udevadm's query and trigger verb now optionally take a .device unit
11619 * systemd-udevd's network naming logic now understands a new
11620 net.naming_scheme= kernel command line switch, which may be used to
11621 pick a specific version of the naming scheme. This helps stabilizing
11622 interface names even as systemd/udev are updated and the naming logic
11625 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
11626 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
11627 initialize one to all 0xFF.
11629 * After loading the SELinux policy systemd will now recursively relabel
11630 all files and directories listed in
11631 /run/systemd/relabel-extra.d/*.relabel (which should be simple
11632 newline separated lists of paths) in addition to the ones it already
11633 implicitly relabels in /run, /dev and /sys. After the relabelling is
11634 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
11635 removed. This is useful to permit initrds (i.e. code running before
11636 the SELinux policy is in effect) to generate files in the host
11637 filesystem safely and ensure that the correct label is applied during
11638 the transition to the host OS.
11640 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
11641 mknod() handling in user namespaces. Previously mknod() would always
11642 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
11643 but device nodes generated that way cannot be opened, and attempts to
11644 open them result in EPERM. This breaks the "graceful fallback" logic
11645 in systemd's PrivateDevices= sand-boxing option. This option is
11646 implemented defensively, so that when systemd detects it runs in a
11647 restricted environment (such as a user namespace, or an environment
11648 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
11649 where device nodes cannot be created the effect of PrivateDevices= is
11650 bypassed (following the logic that 2nd-level sand-boxing is not
11651 essential if the system systemd runs in is itself already sand-boxed
11652 as a whole). This logic breaks with 4.18 in container managers where
11653 user namespacing is used: suddenly PrivateDevices= succeeds setting
11654 up a private /dev/ file system containing devices nodes — but when
11655 these are opened they don't work.
11657 At this point it is recommended that container managers utilizing
11658 user namespaces that intend to run systemd in the payload explicitly
11659 block mknod() with seccomp or similar, so that the graceful fallback
11662 We are very sorry for the breakage and the requirement to change
11663 container configurations for newer kernels. It's purely caused by an
11664 incompatible kernel change. The relevant kernel developers have been
11665 notified about this userspace breakage quickly, but they chose to
11668 * PermissionsStartOnly= setting is deprecated (but is still supported
11669 for backwards compatibility). The same functionality is provided by
11670 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
11673 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
11674 pam_systemd anymore.
11676 * The naming scheme for network devices was changed to always rename
11677 devices, even if they were already renamed by userspace. The "kernel"
11678 policy was changed to only apply as a fallback, if no other naming
11679 policy took effect.
11681 * The requirements to build systemd is bumped to meson-0.46 and
11684 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
11685 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
11686 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
11687 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
11688 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
11689 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
11690 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
11691 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
11692 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
11693 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
11694 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
11695 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
11696 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
11697 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
11698 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
11699 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
11700 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
11701 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
11702 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
11703 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
11704 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
11705 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
11706 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
11707 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
11708 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
11709 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
11710 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
11711 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
11712 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
11713 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
11714 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
11715 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
11716 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
11717 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
11718 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
11719 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
11720 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
11721 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
11722 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
11723 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
11724 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
11725 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
11726 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
11727 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
11728 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
11730 — Warsaw, 2018-12-21
11734 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
11735 builtin will name network interfaces differently than in previous
11736 versions for virtual network interfaces created with SR-IOV and NPAR
11737 and for devices where the PCI network controller device does not have
11738 a slot number associated.
11740 SR-IOV virtual devices are now named based on the name of the parent
11741 interface, with a suffix of "v<N>", where <N> is the virtual device
11742 number. Previously those virtual devices were named as if completely
11745 The ninth and later NPAR virtual devices will be named following the
11746 scheme used for the first eight NPAR partitions. Previously those
11747 devices were not renamed and the kernel default (eth<n>) was used.
11749 "net_id" will also generate names for PCI devices where the PCI
11750 network controller device does not have an associated slot number
11751 itself, but one of its parents does. Previously those devices were
11752 not renamed and the kernel default (eth<n>) was used.
11754 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
11755 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
11756 the unit. So, it is expected that the default behavior of
11757 systemd-logind is not changed. However, if distribution packagers or
11758 administrators disabled or modified IPAddressDeny= setting by a
11759 drop-in config file, then it may be necessary to update the file to
11760 re-enable AF_INET and AF_INET6 to support network user name services,
11763 * When the RestrictNamespaces= unit property is specified multiple
11764 times, then the specified types are merged now. Previously, only the
11765 last assignment was used. So, if distribution packagers or
11766 administrators modified the setting by a drop-in config file, then it
11767 may be necessary to update the file.
11769 * When OnFailure= is used in combination with Restart= on a service
11770 unit, then the specified units will no longer be triggered on
11771 failures that result in restarting. Previously, the specified units
11772 would be activated each time the unit failed, even when the unit was
11773 going to be restarted automatically. This behaviour contradicted the
11774 documentation. With this release the code is adjusted to match the
11777 * systemd-tmpfiles will now print a notice whenever it encounters
11778 tmpfiles.d/ lines referencing the /var/run/ directory. It will
11779 recommend reworking them to use the /run/ directory instead (for
11780 which /var/run/ is simply a symlinked compatibility alias). This way
11781 systemd-tmpfiles can properly detect line conflicts and merge lines
11782 referencing the same file by two paths, without having to access
11785 * systemctl disable/unmask/preset/preset-all cannot be used with
11786 --runtime. Previously this was allowed, but resulted in unintuitive
11787 behaviour that wasn't useful. systemctl disable/unmask will now undo
11788 both runtime and persistent enablement/masking, i.e. it will remove
11789 any relevant symlinks both in /run and /etc.
11791 * Note that all long-running system services shipped with systemd will
11792 now default to a system call allow list (rather than a deny list, as
11793 before). In particular, systemd-udevd will now enforce one too. For
11794 most cases this should be safe, however downstream distributions
11795 which disabled sandboxing of systemd-udevd (specifically the
11796 MountFlags= setting), might want to disable this security feature
11797 too, as the default allow-listing will prohibit all mount, swap,
11798 reboot and clock changing operations from udev rules.
11800 * sd-boot acquired new loader configuration settings to optionally turn
11801 off Windows and MacOS boot partition discovery as well as
11802 reboot-into-firmware menu items. It is also able to pick a better
11803 screen resolution for HiDPI systems, and now provides loader
11804 configuration settings to change the resolution explicitly.
11806 * systemd-resolved now supports DNS-over-TLS. It's still
11807 turned off by default, use DNSOverTLS=opportunistic to turn it on in
11808 resolved.conf. We intend to make this the default as soon as couple
11809 of additional techniques for optimizing the initial latency caused by
11810 establishing a TLS/TCP connection are implemented.
11812 * systemd-resolved.service and systemd-networkd.service now set
11813 DynamicUser=yes. The users systemd-resolve and systemd-network are
11814 not created by systemd-sysusers anymore.
11816 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
11817 that embed a network facing module into any process using getpwuid()
11818 or related call: the dynamic allocation of the user ID for
11819 systemd-resolved.service means the service manager has to check NSS
11820 if the user name is already taken when forking off the service. Since
11821 the user in the common case won't be defined in /etc/passwd the
11822 lookup is likely to trigger nss-ldap which in turn might use NSS to
11823 ask systemd-resolved for hostname lookups. This will hence result in
11824 a deadlock: a user name lookup in order to start
11825 systemd-resolved.service will result in a hostname lookup for which
11826 systemd-resolved.service needs to be started already. There are
11827 multiple ways to work around this problem: pre-allocate the
11828 "systemd-resolve" user on such systems, so that nss-ldap won't be
11829 triggered; or use a different NSS package that doesn't do networking
11830 in-process but provides a local asynchronous name cache; or configure
11831 the NSS package to avoid lookups for UIDs in the range between the
11832 values returned by the commands
11833 'pkg-config systemd --variable=dynamicuidmin' and
11834 'pkg-config systemd --variable=dynamicuidmax', so that it does not
11835 consider itself authoritative for the same UID range systemd
11836 allocates dynamic users from.
11838 * The systemd-resolve tool has been renamed to resolvectl (it also
11839 remains available under the old name, for compatibility), and its
11840 interface is now verb-based, similar in style to the other <xyz>ctl
11841 tools, such as systemctl or loginctl.
11843 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
11844 compatibility. It may be symlinked under the 'resolvconf' name, in
11845 which case it will take arguments and input compatible with the
11846 Debian and FreeBSD resolvconf tool.
11848 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
11849 where the system initially suspends, and after a timeout resumes and
11852 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
11853 set the client will only send a DUID as client identifier. (EDIT: the
11854 option was broken, and was dropped in v255.)
11856 * The nss-systemd glibc NSS module will now enumerate dynamic users and
11857 groups in effect. Previously, it could resolve UIDs/GIDs to user
11858 names/groups and vice versa, but did not support enumeration.
11860 * journald's Compress= configuration setting now optionally accepts a
11861 byte threshold value. All journal objects larger than this threshold
11862 will be compressed, smaller ones will not. Previously this threshold
11863 was not configurable and set to 512.
11865 * A new system.conf setting NoNewPrivileges= is now available which may
11866 be used to turn off acquisition of new privileges system-wide
11867 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
11868 for all its children). Note that turning this option on means setuid
11869 binaries and file system capabilities lose their special powers.
11870 While turning on this option is a big step towards a more secure
11871 system, doing so is likely to break numerous pre-existing UNIX tools,
11872 in particular su and sudo.
11874 * A new service systemd-time-sync-wait.service has been added. If
11875 enabled it will delay the time-sync.target unit at boot until time
11876 synchronization has been received from the network. This
11877 functionality is useful on systems lacking a local RTC or where it is
11878 acceptable that the boot process shall be delayed by external network
11881 * When hibernating, systemd will now inform the kernel of the image
11882 write offset, on kernels new enough to support this. This means swap
11883 files should work for hibernation now.
11885 * When loading unit files, systemd will now look for drop-in unit files
11886 extensions in additional places. Previously, for a unit file name
11887 "foo-bar-baz.service" it would look for dropin files in
11888 "foo-bar-baz.service.d/*.conf". Now, it will also look in
11889 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
11890 service name truncated after all inner dashes. This scheme allows
11891 writing drop-ins easily that apply to a whole set of unit files at
11892 once. It's particularly useful for mount and slice units (as their
11893 naming is prefix based), but is also useful for service and other
11894 units, for packages that install multiple unit files at once,
11895 following a strict naming regime of beginning the unit file name with
11896 the package's name. Two new specifiers are now supported in unit
11897 files to match this: %j and %J are replaced by the part of the unit
11898 name following the last dash.
11900 * Unit files and other configuration files that support specifier
11901 expansion now understand another three new specifiers: %T and %V will
11902 resolve to /tmp and /var/tmp respectively, or whatever temporary
11903 directory has been set for the calling user. %E will expand to either
11904 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
11906 * The ExecStart= lines of unit files are no longer required to
11907 reference absolute paths. If non-absolute paths are specified the
11908 specified binary name is searched within the service manager's
11909 built-in $PATH, which may be queried with 'systemd-path
11910 search-binaries-default'. It's generally recommended to continue to
11911 use absolute paths for all binaries specified in unit files.
11913 * Units gained a new load state "bad-setting", which is used when a
11914 unit file was loaded, but contained fatal errors which prevent it
11915 from being started (for example, a service unit has been defined
11916 lacking both ExecStart= and ExecStop= lines).
11918 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
11919 support alternative debuggers, for example lldb. The old name
11920 continues to be available however, for compatibility reasons. Use the
11921 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
11922 to pick an alternative debugger instead of the default gdb.
11924 * systemctl and the other tools will now output escape sequences that
11925 generate proper clickable hyperlinks in various terminal emulators
11926 where useful (for example, in the "systemctl status" output you can
11927 now click on the unit file name to quickly open it in the
11928 editor/viewer of your choice). Note that not all terminal emulators
11929 support this functionality yet, but many do. Unfortunately, the
11930 "less" pager doesn't support this yet, hence this functionality is
11931 currently automatically turned off when a pager is started (which
11932 happens quite often due to auto-paging). We hope to remove this
11933 limitation as soon as "less" learns these escape sequences. This new
11934 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
11935 environment variable. For details on these escape sequences see:
11936 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
11938 * networkd's .network files now support a new IPv6MTUBytes= option for
11939 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
11940 option in the [Route] section to configure the MTU to use for
11941 specific routes. It also gained support for configuration of the DHCP
11942 "UserClass" option through the new UserClass= setting. It gained
11943 three new options in the new [CAN] section for configuring CAN
11944 networks. The MULTICAST and ALLMULTI interface flags may now be
11945 controlled explicitly with the new Multicast= and AllMulticast=
11948 * networkd will now automatically make use of the kernel's route
11949 expiration feature, if it is available.
11951 * udevd's .link files now support setting the number of receive and
11952 transmit channels, using the RxChannels=, TxChannels=,
11953 OtherChannels=, CombinedChannels= settings.
11955 * Support for UDPSegmentationOffload= has been removed, given its
11956 limited support in hardware, and waning software support.
11958 * networkd's .netdev files now support creating "netdevsim" interfaces.
11960 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
11961 to query the unit belonging to a specific kernel control group.
11963 * systemd-analyze gained a new verb "cat-config", which may be used to
11964 dump the contents of any configuration file, with all its matching
11965 drop-in files added in, and honouring the usual search and masking
11966 logic applied to systemd configuration files. For example use
11967 "systemd-analyze cat-config systemd/system.conf" to get the complete
11968 system configuration file of systemd how it would be loaded by PID 1
11969 itself. Similar to this, various tools such as systemd-tmpfiles or
11970 systemd-sysusers, gained a new option "--cat-config", which does the
11971 corresponding operation for their own configuration settings. For
11972 example, "systemd-tmpfiles --cat-config" will now output the full
11973 list of tmpfiles.d/ lines in place.
11975 * timedatectl gained three new verbs: "show" shows bus properties of
11976 systemd-timedated, "timesync-status" shows the current NTP
11977 synchronization state of systemd-timesyncd, and "show-timesync"
11978 shows bus properties of systemd-timesyncd.
11980 * systemd-timesyncd gained a bus interface on which it exposes details
11983 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
11984 understood by systemd-timedated. It takes a colon-separated list of
11985 unit names of NTP client services. The list is used by
11986 "timedatectl set-ntp".
11988 * systemd-nspawn gained a new --rlimit= switch for setting initial
11989 resource limits for the container payload. There's a new switch
11990 --hostname= to explicitly override the container's hostname. A new
11991 --no-new-privileges= switch may be used to control the
11992 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
11993 --oom-score-adjust= switch controls the OOM scoring adjustment value
11994 for the payload. The new --cpu-affinity= switch controls the CPU
11995 affinity of the container payload. The new --resolv-conf= switch
11996 allows more detailed control of /etc/resolv.conf handling of the
11997 container. Similarly, the new --timezone= switch allows more detailed
11998 control of /etc/localtime handling of the container.
12000 * systemd-detect-virt gained a new --list switch, which will print a
12001 list of all currently known VM and container environments.
12003 * Support for "Portable Services" has been added, see
12004 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
12005 experimental, but this is expected to change soon. Reflecting this
12006 experimental state, the "portablectl" binary is not installed into
12007 /usr/bin yet. The binary has to be called with the full path
12008 /usr/lib/systemd/portablectl instead.
12010 * journalctl's and systemctl's -o switch now knows a new log output
12011 mode "with-unit". The output it generates is very similar to the
12012 regular "short" mode, but displays the unit name instead of the
12013 syslog tag for each log line. Also, the date is shown with timezone
12014 information. This mode is probably more useful than the classic
12015 "short" output mode for most purposes, except where pixel-perfect
12016 compatibility with classic /var/log/messages formatting is required.
12018 * A new --dump-bus-properties switch has been added to the systemd
12019 binary, which may be used to dump all supported D-Bus properties.
12020 (Options which are still supported, but are deprecated, are *not*
12023 * sd-bus gained a set of new calls:
12024 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
12025 enable/disable the "floating" state of a bus slot object,
12026 i.e. whether the slot object pins the bus it is allocated for into
12027 memory or if the bus slot object gets disconnected when the bus goes
12028 away. sd_bus_open_with_description(),
12029 sd_bus_open_user_with_description(),
12030 sd_bus_open_system_with_description() may be used to allocate bus
12031 objects and set their description string already during allocation.
12033 * sd-event gained support for watching inotify events from the event
12034 loop, in an efficient way, sharing inotify handles between multiple
12035 users. For this a new function sd_event_add_inotify() has been added.
12037 * sd-event and sd-bus gained support for calling special user-supplied
12038 destructor functions for userdata pointers associated with
12039 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
12040 functions sd_bus_slot_set_destroy_callback,
12041 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
12042 sd_bus_track_get_destroy_callback,
12043 sd_event_source_set_destroy_callback,
12044 sd_event_source_get_destroy_callback have been added.
12046 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
12048 * PID 1 will now automatically reschedule .timer units whenever the
12049 local timezone changes. (They previously got rescheduled
12050 automatically when the system clock changed.)
12052 * New documentation has been added to document cgroups delegation,
12053 portable services and the various code quality tools we have set up:
12055 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
12056 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
12057 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
12059 * The Boot Loader Specification has been added to the source tree.
12061 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
12063 While moving it into our source tree we have updated it and further
12064 changes are now accepted through the usual github PR workflow.
12066 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
12067 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
12068 earlier PAM modules. The data in these fields is used to initialize
12069 the session scope's resource properties. Thus external PAM modules
12070 may now configure per-session limits, for example sourced from
12071 external user databases.
12073 * socket units with Accept=yes will now maintain a "refused" counter in
12074 addition to the existing "accepted" counter, counting connections
12075 refused due to the enforced limits.
12077 * The "systemd-path search-binaries-default" command may now be use to
12078 query the default, built-in $PATH PID 1 will pass to the services it
12081 * A new unit file setting PrivateMounts= has been added. It's a boolean
12082 option. If enabled the unit's processes are invoked in their own file
12083 system namespace. Note that this behaviour is also implied if any
12084 other file system namespacing options (such as PrivateTmp=,
12085 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
12086 primarily useful for services that do not use any of the other file
12087 system namespacing options. One such service is systemd-udevd.service
12088 where this is now used by default.
12090 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
12091 when the system is booted in UEFI "secure mode".
12093 * A new unit "system-update-pre.target" is added, which defines an
12094 optional synchronization point for offline system updates, as
12095 implemented by the pre-existing "system-update.target" unit. It
12096 allows ordering services before the service that executes the actual
12097 update process in a generic way.
12099 * Systemd now emits warnings whenever .include syntax is used.
12101 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
12102 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
12103 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
12104 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
12105 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
12106 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
12107 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
12108 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
12109 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
12110 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
12111 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
12112 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
12113 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
12114 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
12115 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
12116 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
12117 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
12118 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
12119 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
12120 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
12121 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
12122 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
12123 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
12124 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
12125 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
12126 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
12127 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
12128 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
12129 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
12131 — Berlin, 2018-06-22
12135 * The MemoryAccounting= unit property now defaults to on. After
12136 discussions with the upstream control group maintainers we learnt
12137 that the negative impact of cgroup memory accounting on current
12138 kernels is finally relatively minimal, so that it should be safe to
12139 enable this by default without affecting system performance. Besides
12140 memory accounting only task accounting is turned on by default, all
12141 other forms of resource accounting (CPU, IO, IP) remain off for now,
12142 because it's not clear yet that their impact is small enough to move
12143 from opt-in to opt-out. We recommend downstreams to leave memory
12144 accounting on by default if kernel 4.14 or higher is primarily
12145 used. On very resource constrained systems or when support for old
12146 kernels is a necessity, -Dmemory-accounting-default=false can be used
12147 to revert this change.
12149 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
12150 %udev_rules_update) and the journal catalog (%journal_catalog_update)
12151 from the upgrade scriptlets of individual packages now do nothing.
12152 Transfiletriggers have been added which will perform those updates
12153 once at the end of the transaction.
12155 Similar transfiletriggers have been added to execute any sysctl.d
12156 and binfmt.d rules. Thus, it should be unnecessary to provide any
12157 scriptlets to execute this configuration from package installation
12160 * systemd-sysusers gained a mode where the configuration to execute is
12161 specified on the command line, but this configuration is not executed
12162 directly, but instead it is merged with the configuration on disk,
12163 and the result is executed. This is useful for package installation
12164 scripts which want to create the user before installing any files on
12165 disk (in case some of those files are owned by that user), while
12166 still allowing local admin overrides.
12168 This functionality is exposed to rpm scriptlets through a new
12169 %sysusers_create_package macro. Old %sysusers_create and
12170 %sysusers_create_inline macros are deprecated.
12172 A transfiletrigger for sysusers.d configuration is now installed,
12173 which means that it should be unnecessary to call systemd-sysusers from
12174 package installation scripts, unless the package installs any files
12175 owned by those newly-created users, in which case
12176 %sysusers_create_package should be used.
12178 * Analogous change has been done for systemd-tmpfiles: it gained a mode
12179 where the command-line configuration is merged with the configuration
12180 on disk. This is exposed as the new %tmpfiles_create_package macro,
12181 and %tmpfiles_create is deprecated. A transfiletrigger is installed
12182 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
12183 from package installation scripts.
12185 * sysusers.d configuration for a user may now also specify the group
12186 number, in addition to the user number ("u username 123:456"), or
12187 without the user number ("u username -:456").
12189 * Configution items for systemd-sysusers can now be specified as
12190 positional arguments when the new --inline switch is used.
12192 * The login shell of users created through sysusers.d may now be
12193 specified (previously, it was always /bin/sh for root and
12194 /sbin/nologin for other users).
12196 * systemd-analyze gained a new --global switch to look at global user
12197 configuration. It also gained a unit-paths verb to list the unit load
12198 paths that are compiled into systemd (which can be used with
12199 --systemd, --user, or --global).
12201 * udevadm trigger gained a new --settle/-w option to wait for any
12202 triggered events to finish (but just those, and not any other events
12203 which are triggered meanwhile).
12205 * The action that systemd-logind takes when the lid is closed and the
12206 machine is connected to external power can now be configured using
12207 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
12208 was determined by HandleLidSwitch=, and, for backwards compatibility,
12209 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
12211 * journalctl will periodically call sd_journal_process() to make it
12212 resilient against inotify queue overruns when journal files are
12213 rotated very quickly.
12215 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
12216 sd_bus_get_n_queued_write — may be used to check the number of
12217 pending bus messages.
12219 * systemd gained a new
12220 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
12221 which can be used to migrate foreign processes to scope and service
12222 units. The primary user for this new API is systemd itself: the
12223 systemd --user instance uses this call of the systemd --system
12224 instance to migrate processes if it itself gets the request to
12225 migrate processes and the kernel refuses this due to access
12226 restrictions. Thanks to this "systemd-run --scope --user …" works
12227 again in pure cgroup v2 environments when invoked from the user
12230 * A new TemporaryFileSystem= setting can be used to mask out part of
12231 the real file system tree with tmpfs mounts. This may be combined
12232 with BindPaths= and BindReadOnlyPaths= to hide files or directories
12233 not relevant to the unit, while still allowing some paths lower in
12234 the tree to be accessed.
12236 ProtectHome=tmpfs may now be used to hide user home and runtime
12237 directories from units, in a way that is mostly equivalent to
12238 "TemporaryFileSystem=/home /run/user /root".
12240 * Non-service units are now started with KeyringMode=shared by default.
12241 This means that mount and swapon and other mount tools have access
12242 to keys in the main keyring.
12244 * /sys/fs/bpf is now mounted automatically.
12246 * QNX virtualization is now detected by systemd-detect-virt and may
12247 be used in ConditionVirtualization=.
12249 * IPAccounting= may now be enabled also for slice units.
12251 * A new -Dsplit-bin= build configuration switch may be used to specify
12252 whether bin and sbin directories are merged, or if they should be
12253 included separately in $PATH and various listings of executable
12254 directories. The build configuration scripts will try to autodetect
12255 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
12256 system, but distributions are encouraged to configure this
12259 * A new -Dok-color= build configuration switch may be used to change
12260 the colour of "OK" status messages.
12262 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
12263 PrivateNetwork=yes was buggy in previous versions of systemd. This
12264 means that after the upgrade and daemon-reexec, any such units must
12267 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
12268 will not exclude read-only files owned by root from cleanup.
12270 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
12271 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
12272 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
12273 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
12274 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
12275 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
12276 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
12277 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
12278 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
12279 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
12280 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
12281 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
12282 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
12283 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
12284 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
12285 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
12287 — Warsaw, 2018-03-05
12291 * Some keyboards come with a zoom see-saw or rocker which until now got
12292 mapped to the Linux "zoomin/out" keys in hwdb. However, these
12293 keycodes are not recognized by any major desktop. They now produce
12294 Up/Down key events so that they can be used for scrolling.
12296 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
12297 slightly: previously, if an argument was specified for lines of this
12298 type (i.e. the right-most column was set) this string was appended to
12299 existing files each time systemd-tmpfiles was run. This behaviour was
12300 different from what the documentation said, and not particularly
12301 useful, as repeated systemd-tmpfiles invocations would not be
12302 idempotent and grow such files without bounds. With this release
12303 behaviour has been altered to match what the documentation says:
12304 lines of this type only have an effect if the indicated files don't
12305 exist yet, and only then the argument string is written to the file.
12307 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
12308 systemd-tmpfiles behaviour: previously, read-only files owned by root
12309 were always excluded from the file "aging" algorithm (i.e. the
12310 automatic clean-up of directories like /tmp based on
12311 atime/mtime/ctime). We intend to drop this restriction, and age files
12312 by default even when owned by root and read-only. This behaviour was
12313 inherited from older tools, but there have been requests to remove
12314 it, and it's not obvious why this restriction was made in the first
12315 place. Please speak up now, if you are aware of software that requires
12316 this behaviour, otherwise we'll remove the restriction in v238.
12318 * A new environment variable $SYSTEMD_OFFLINE is now understood by
12319 systemctl. It takes a boolean argument. If on, systemctl assumes it
12320 operates on an "offline" OS tree, and will not attempt to talk to the
12321 service manager. Previously, this mode was implicitly enabled if a
12322 chroot() environment was detected, and this new environment variable
12323 now provides explicit control.
12325 * .path and .socket units may now be created transiently, too.
12326 Previously only service, mount, automount and timer units were
12327 supported as transient units. The systemd-run tool has been updated
12328 to expose this new functionality, you may hence use it now to bind
12329 arbitrary commands to path or socket activation on-the-fly from the
12330 command line. Moreover, almost all properties are now exposed for the
12331 unit types that already supported transient operation.
12333 * The systemd-mount command gained support for a new --owner= parameter
12334 which takes a user name, which is then resolved and included in uid=
12335 and gid= mount options string of the file system to mount.
12337 * A new unit condition ConditionControlGroupController= has been added
12338 that checks whether a specific cgroup controller is available.
12340 * Unit files, udev's .link files, and systemd-networkd's .netdev and
12341 .network files all gained support for a new condition
12342 ConditionKernelVersion= for checking against specific kernel
12345 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
12346 support for configuring device flags in the Flags= setting. In the
12347 same files, the [Tunnel] section gained support for configuring
12348 AllowLocalRemote=. The [Route] section in .network files gained
12349 support for configuring InitialCongestionWindow=,
12350 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
12351 understands RapidCommit=.
12353 * systemd-networkd's DHCPv6 support gained support for Prefix
12356 * sd-bus gained support for a new "watch-bind" feature. When this
12357 feature is enabled, an sd_bus connection may be set up to connect to
12358 an AF_UNIX socket in the file system as soon as it is created. This
12359 functionality is useful for writing early-boot services that
12360 automatically connect to the system bus as soon as it is started,
12361 without ugly time-based polling. systemd-networkd and
12362 systemd-resolved have been updated to make use of this
12363 functionality. busctl exposes this functionality in a new
12364 --watch-bind= command line switch.
12366 * sd-bus will now optionally synthesize a local "Connected" signal as
12367 soon as a D-Bus connection is set up fully. This message mirrors the
12368 already existing "Disconnected" signal which is synthesized when the
12369 connection is terminated. This signal is generally useful but
12370 particularly handy in combination with the "watch-bind" feature
12371 described above. Synthesizing of this message has to be requested
12372 explicitly through the new API call sd_bus_set_connected_signal(). In
12373 addition a new call sd_bus_is_ready() has been added that checks
12374 whether a connection is fully set up (i.e. between the "Connected" and
12375 "Disconnected" signals).
12377 * sd-bus gained two new calls sd_bus_request_name_async() and
12378 sd_bus_release_name_async() for asynchronously registering bus
12379 names. Similar, there is now sd_bus_add_match_async() for installing
12380 a signal match asynchronously. All of systemd's own services have
12381 been updated to make use of these calls. Doing these operations
12382 asynchronously has two benefits: it reduces the risk of deadlocks in
12383 case of cyclic dependencies between bus services, and it speeds up
12384 service initialization since synchronization points for bus
12385 round-trips are removed.
12387 * sd-bus gained two new calls sd_bus_match_signal() and
12388 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
12389 and sd_bus_add_match_async() but instead of taking a D-Bus match
12390 string take match fields as normal function parameters.
12392 * sd-bus gained two new calls sd_bus_set_sender() and
12393 sd_bus_message_set_sender() for setting the sender name of outgoing
12394 messages (either for all outgoing messages or for just one specific
12395 one). These calls are only useful in direct connections as on
12396 brokered connections the broker fills in the sender anyway,
12397 overwriting whatever the client filled in.
12399 * sd-event gained a new pseudo-handle that may be specified on all API
12400 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
12401 used this refers to the default event loop object of the calling
12402 thread. Note however that this does not implicitly allocate one —
12403 which has to be done prior by using sd_event_default(). Similarly
12404 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
12405 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
12406 to the default bus of the specified type of the calling thread. Here
12407 too this does not implicitly allocate bus connection objects, this
12408 has to be done prior with sd_bus_default() and friends.
12410 * sd-event gained a new call pair
12411 sd_event_source_{get|set}_io_fd_own(). This may be used to request
12412 automatic closure of the file descriptor an IO event source watches
12413 when the event source is destroyed.
12415 * systemd-networkd gained support for natively configuring WireGuard
12418 * In previous versions systemd synthesized user records both for the
12419 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
12420 internally. In order to simplify distribution-wide renames of the
12421 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
12422 new transitional flag file has been added: if
12423 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
12424 user and group record within the systemd codebase is disabled.
12426 * systemd-notify gained a new --uid= option for selecting the source
12427 user/UID to use for notification messages sent to the service
12430 * journalctl gained a new --grep= option to list only entries in which
12431 the message matches a certain pattern. By default matching is case
12432 insensitive if the pattern is lowercase, and case sensitive
12433 otherwise. Option --case-sensitive=yes|no can be used to override
12434 this an specify case sensitivity or case insensitivity.
12436 * There's now a "systemd-analyze service-watchdogs" command for printing
12437 the current state of the service runtime watchdog, and optionally
12438 enabling or disabling the per-service watchdogs system-wide if given a
12439 boolean argument (i.e. the concept you configure in WatchdogSec=), for
12440 debugging purposes. There's also a kernel command line option
12441 systemd.service_watchdogs= for controlling the same.
12443 * Two new "log-level" and "log-target" options for systemd-analyze were
12444 added that merge the now deprecated get-log-level, set-log-level and
12445 get-log-target, set-log-target pairs. The deprecated options are still
12446 understood for backwards compatibility. The two new options print the
12447 current value when no arguments are given, and set them when a
12448 level/target is given as an argument.
12450 * sysusers.d's "u" lines now optionally accept both a UID and a GID
12451 specification, separated by a ":" character, in order to create users
12452 where UID and GID do not match.
12454 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
12455 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
12456 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
12457 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
12458 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
12459 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
12460 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
12461 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
12462 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
12463 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
12464 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
12465 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
12466 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
12467 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
12468 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
12469 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
12470 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
12471 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
12472 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
12473 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
12480 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
12481 in v235 has been extended to also set the dummy.ko module option
12482 numdummies=0, preventing the kernel from automatically creating
12483 dummy0. All dummy interfaces must now be explicitly created.
12485 * Unknown '%' specifiers in configuration files are now rejected. This
12486 applies to units and tmpfiles.d configuration. Any percent characters
12487 that are followed by a letter or digit that are not supposed to be
12488 interpreted as the beginning of a specifier should be escaped by
12489 doubling ("%%"). (So "size=5%" is still accepted, as well as
12490 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
12491 valid specifiers today.)
12493 * systemd-resolved now maintains a new dynamic
12494 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
12495 recommended to make /etc/resolv.conf a symlink to it. This file
12496 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
12497 includes dynamically acquired search domains, achieving more correct
12498 DNS resolution by software that bypasses local DNS APIs such as NSS.
12500 * The "uaccess" udev tag has been dropped from /dev/kvm and
12501 /dev/dri/renderD*. These devices now have the 0666 permissions by
12502 default (but this may be changed at build-time). /dev/dri/renderD*
12503 will now be owned by the "render" group along with /dev/kfd.
12505 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
12506 systemd-journal-gatewayd.service and
12507 systemd-journal-upload.service. This means "nss-systemd" must be
12508 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
12509 services are resolved properly.
12511 * In /etc/fstab two new mount options are now understood:
12512 x-systemd.makefs and x-systemd.growfs. The former has the effect that
12513 the configured file system is formatted before it is mounted, the
12514 latter that the file system is resized to the full block device size
12515 after it is mounted (i.e. if the file system is smaller than the
12516 partition it resides on, it's grown). This is similar to the fsck
12517 logic in /etc/fstab, and pulls in systemd-makefs@.service and
12518 systemd-growfs@.service as necessary, similar to
12519 systemd-fsck@.service. Resizing is currently only supported on ext4
12522 * In systemd-networkd, the IPv6 RA logic now optionally may announce
12523 DNS server and domain information.
12525 * Support for the LUKS2 on-disk format for encrypted partitions has
12526 been added. This requires libcryptsetup2 during compilation and
12529 * The systemd --user instance will now signal "readiness" when its
12530 basic.target unit has been reached, instead of when the run queue ran
12531 empty for the first time.
12533 * Tmpfiles.d with user configuration are now also supported.
12534 systemd-tmpfiles gained a new --user switch, and snippets placed in
12535 ~/.config/user-tmpfiles.d/ and corresponding directories will be
12536 executed by systemd-tmpfiles --user running in the new
12537 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
12538 running in the user session.
12540 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
12541 %S resolves to the top-level state directory (/var/lib for the system
12542 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
12543 top-level cache directory (/var/cache for the system instance,
12544 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
12545 logs directory (/var/log for the system instance,
12546 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
12547 existing %t specifier, that resolves to the top-level runtime
12548 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
12551 * journalctl learnt a new parameter --output-fields= for limiting the
12552 set of journal fields to output in verbose and JSON output modes.
12554 * systemd-timesyncd's configuration file gained a new option
12555 RootDistanceMaxSec= for setting the maximum root distance of servers
12556 it'll use, as well as the new options PollIntervalMinSec= and
12557 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
12559 * bootctl gained a new command "list" for listing all available boot
12560 menu items on systems that follow the boot loader specification.
12562 * systemctl gained a new --dry-run switch that shows what would be done
12563 instead of doing it, and is currently supported by the shutdown and
12566 * ConditionSecurity= can now detect the TOMOYO security module.
12568 * Unit file [Install] sections are now also respected in unit drop-in
12569 files. This is intended to be used by drop-ins under /usr/lib/.
12571 * systemd-firstboot may now also set the initial keyboard mapping.
12573 * Udev "changed" events for devices which are exposed as systemd
12574 .device units are now propagated to units specified in
12575 ReloadPropagatedFrom= as reload requests.
12577 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
12578 unit template name (i.e. a name in the form of 'foobar@.service',
12579 without the instance component between the '@' and - the '.'), then
12580 the escaped sysfs path of the device is automatically used as the
12583 * SystemCallFilter= in unit files has been extended so that an "errno"
12584 can be specified individually for each system call. Example:
12585 SystemCallFilter=~uname:EILSEQ.
12587 * The cgroup delegation logic has been substantially updated. Delegate=
12588 now optionally takes a list of controllers (instead of a boolean, as
12589 before), which lists the controllers to delegate at least.
12591 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
12593 * A new LogLevelMax= setting configures the maximum log level any
12594 process of the service may log at (i.e. anything with a lesser
12595 priority than what is specified is automatically dropped). A new
12596 LogExtraFields= setting allows configuration of additional journal
12597 fields to attach to all log records generated by any of the unit's
12600 * New StandardInputData= and StandardInputText= settings along with the
12601 new option StandardInput=data may be used to configure textual or
12602 binary data that shall be passed to the executed service process via
12603 standard input, encoded in-line in the unit file.
12605 * StandardInput=, StandardOutput= and StandardError= may now be used to
12606 connect stdin/stdout/stderr of executed processes directly with a
12607 file or AF_UNIX socket in the file system, using the new "file:" option.
12609 * A new unit file option CollectMode= has been added, that allows
12610 tweaking the garbage collection logic for units. It may be used to
12611 tell systemd to garbage collect units that have failed automatically
12612 (normally it only GCs units that exited successfully). systemd-run
12613 and systemd-mount expose this new functionality with a new -G option.
12615 * "machinectl bind" may now be used to bind mount non-directories
12616 (i.e. regularfiles, devices, fifos, sockets).
12618 * systemd-analyze gained a new verb "calendar" for validating and
12619 testing calendar time specifications to use for OnCalendar= in timer
12620 units. Besides validating the expression it will calculate the next
12621 time the specified expression would elapse.
12623 * In addition to the pre-existing FailureAction= unit file setting
12624 there's now SuccessAction=, for configuring a shutdown action to
12625 execute when a unit completes successfully. This is useful in
12626 particular inside containers that shall terminate after some workload
12627 has been completed. Also, both options are now supported for all unit
12628 types, not just services.
12630 * networkds's IP rule support gained two new options
12631 IncomingInterface= and OutgoingInterface= for configuring the incoming
12632 and outgoing interfaces of configured rules. systemd-networkd also
12633 gained support for "vxcan" network devices.
12635 * networkd gained a new setting RequiredForOnline=, taking a
12636 boolean. If set, systemd-wait-online will take it into consideration
12637 when determining that the system is up, otherwise it will ignore the
12638 interface for this purpose.
12640 * The sd_notify() protocol gained support for a new operation: with
12641 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
12642 store again, ahead of POLLHUP or POLLERR when they are removed
12645 * A new document doc/UIDS-GIDS.md has been added to the source tree,
12646 that documents the UID/GID range and assignment assumptions and
12647 requirements of systemd.
12649 * The watchdog device PID 1 will ping may now be configured through the
12650 WatchdogDevice= configuration file setting, or by setting the
12651 systemd.watchdog_service= kernel command line option.
12653 * systemd-resolved's gained support for registering DNS-SD services on
12654 the local network using MulticastDNS. Services may either be
12655 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
12656 the same dir below /run, /usr/lib), or through its D-Bus API.
12658 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
12659 extend the effective start, runtime, and stop time. The service must
12660 continue to send EXTEND_TIMEOUT_USEC within the period specified to
12661 prevent the service manager from making the service as timedout.
12663 * systemd-resolved's DNSSEC support gained support for RFC 8080
12664 (Ed25519 keys and signatures).
12666 * The systemd-resolve command line tool gained a new set of options
12667 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
12668 --set-nta= and --revert to configure per-interface DNS configuration
12669 dynamically during runtime. It's useful for pushing DNS information
12670 into systemd-resolved from DNS hook scripts that various interface
12671 managing software supports (such as pppd).
12673 * systemd-nspawn gained a new --network-namespace-path= command line
12674 option, which may be used to make a container join an existing
12675 network namespace, by specifying a path to a "netns" file.
12677 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
12678 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
12679 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
12680 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
12681 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
12682 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
12683 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
12684 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
12685 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
12686 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
12687 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
12688 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
12689 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
12690 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
12691 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
12692 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
12693 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
12694 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
12695 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
12696 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
12697 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
12698 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
12699 Jędrzejewski-Szmek, Zeal Jagannatha
12701 — Berlin, 2017-12-14
12705 * INCOMPATIBILITY: systemd-logind.service and other long-running
12706 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
12707 communication with the outside. This generally improves security of
12708 the system, and is in almost all cases a safe and good choice, as
12709 these services do not and should not provide any network-facing
12710 functionality. However, systemd-logind uses the glibc NSS API to
12711 query the user database. This creates problems on systems where NSS
12712 is set up to directly consult network services for user database
12713 lookups. In particular, this creates incompatibilities with the
12714 "nss-nis" module, which attempts to directly contact the NIS/YP
12715 network servers it is configured for, and will now consistently
12716 fail. In such cases, it is possible to turn off IP sandboxing for
12717 systemd-logind.service (set IPAddressDeny= in its [Service] section
12718 to the empty string, via a .d/ unit file drop-in). Downstream
12719 distributions might want to update their nss-nis packaging to include
12720 such a drop-in snippet, accordingly, to hide this incompatibility
12721 from the user. Another option is to make use of glibc's nscd service
12722 to proxy such network requests through a privilege-separated, minimal
12723 local caching daemon, or to switch to more modern technologies such
12724 sssd, whose NSS hook-ups generally do not involve direct network
12725 access. In general, we think it's definitely time to question the
12726 implementation choices of nss-nis, i.e. whether it's a good idea
12727 today to embed a network-facing loadable module into all local
12728 processes that need to query the user database, including the most
12729 trivial and benign ones, such as "ls". For more details about
12730 IPAddressDeny= see below.
12732 * A new modprobe.d drop-in is now shipped by default that sets the
12733 bonding module option max_bonds=0. This overrides the kernel default,
12734 to avoid conflicts and ambiguity as to whether or not bond0 should be
12735 managed by systemd-networkd or not. This resolves multiple issues
12736 with bond0 properties not being applied, when bond0 is configured
12737 with systemd-networkd. Distributors may choose to not package this,
12738 however in that case users will be prevented from correctly managing
12739 bond0 interface using systemd-networkd.
12741 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
12742 which print the logging level and target of the system manager. They
12743 complement the existing "set-log-level" and "set-log-target" verbs
12744 used to change those values.
12746 * journald.conf gained a new boolean setting ReadKMsg= which defaults
12747 to on. If turned off kernel log messages will not be read by
12748 systemd-journald or included in the logs. It also gained a new
12749 setting LineMax= for configuring the maximum line length in
12750 STDOUT/STDERR log streams. The new default for this value is 48K, up
12751 from the previous hardcoded 2048.
12753 * A new unit setting RuntimeDirectoryPreserve= has been added, which
12754 allows more detailed control of what to do with a runtime directory
12755 configured with RuntimeDirectory= (i.e. a directory below /run or
12756 $XDG_RUNTIME_DIR) after a unit is stopped.
12758 * The RuntimeDirectory= setting for units gained support for creating
12759 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
12760 one top-level directory.
12762 * Units gained new options StateDirectory=, CacheDirectory=,
12763 LogsDirectory= and ConfigurationDirectory= which are closely related
12764 to RuntimeDirectory= but manage per-service directories below
12765 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
12766 possible to write unit files which when activated automatically gain
12767 properly owned service specific directories in these locations, thus
12768 making unit files self-contained and increasing compatibility with
12769 stateless systems and factory reset where /etc or /var are
12770 unpopulated at boot. Matching these new settings there's also
12771 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
12772 ConfigurationDirectoryMode= for configuring the access mode of these
12773 directories. These settings are particularly useful in combination
12774 with DynamicUser=yes as they provide secure, properly-owned,
12775 writable, and stateful locations for storage, excluded from the
12776 sandbox that such services live in otherwise.
12778 * Automake support has been removed from this release. systemd is now
12781 * systemd-journald will now aggressively cache client metadata during
12782 runtime, speeding up log write performance under pressure. This comes
12783 at a small price though: as much of the metadata is read
12784 asynchronously from /proc/ (and isn't implicitly attached to log
12785 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
12786 metadata stored alongside a log entry might be slightly
12787 out-of-date. Previously it could only be slightly newer than the log
12788 message. The time window is small however, and given that the kernel
12789 is unlikely to be improved anytime soon in this regard, this appears
12792 * nss-myhostname/systemd-resolved will now by default synthesize an
12793 A/AAAA resource record for the "_gateway" hostname, pointing to the
12794 current default IP gateway. Previously it did that for the "gateway"
12795 name, hampering adoption, as some distributions wanted to leave that
12796 hostname open for local use. The old behaviour may still be
12797 requested at build time.
12799 * systemd-networkd's [Address] section in .network files gained a new
12800 Scope= setting for configuring the IP address scope. The [Network]
12801 section gained a new boolean setting ConfigureWithoutCarrier= that
12802 tells systemd-networkd to ignore link sensing when configuring the
12803 device. The [DHCP] section gained a new Anonymize= boolean option for
12804 turning on a number of options suggested in RFC 7844. A new
12805 [RoutingPolicyRule] section has been added for configuring the IP
12806 routing policy. The [Route] section has gained support for a new
12807 Type= setting which permits configuring
12808 blackhole/unreachable/prohibit routes.
12810 * The [VRF] section in .netdev files gained a new Table= setting for
12811 configuring the routing table to use. The [Tunnel] section gained a
12812 new Independent= boolean field for configuring tunnels independent of
12813 an underlying network interface. The [Bridge] section gained a new
12814 GroupForwardMask= option for configuration of propagation of link
12815 local frames between bridge ports.
12817 * The WakeOnLan= setting in .link files gained support for a number of
12818 new modes. A new TCP6SegmentationOffload= setting has been added for
12819 configuring TCP/IPv6 hardware segmentation offload.
12821 * The IPv6 RA sender implementation may now optionally send out RDNSS
12822 and RDNSSL records to supply DNS configuration to peers.
12824 * systemd-nspawn gained support for a new --system-call-filter= command
12825 line option for adding and removing entries in the default system
12826 call filter it applies. Moreover systemd-nspawn has been changed to
12827 implement a system call allow list instead of a deny list.
12829 * systemd-run gained support for a new --pipe command line option. If
12830 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
12831 are directly passed on to the activated transient service
12832 executable. This allows invoking arbitrary processes as systemd
12833 services (for example to take benefit of dependency management,
12834 accounting management, resource management or log management that is
12835 done automatically for services) — while still allowing them to be
12836 integrated in a classic UNIX shell pipeline.
12838 * When a service sends RELOAD=1 via sd_notify() and reload propagation
12839 using ReloadPropagationTo= is configured, a reload is now propagated
12840 to configured units. (Previously this was only done on explicitly
12841 requested reloads, using "systemctl reload" or an equivalent
12844 * For each service unit a restart counter is now kept: it is increased
12845 each time the service is restarted due to Restart=, and may be
12846 queried using "systemctl show -p NRestarts …".
12848 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
12849 @signal and @timer have been added, for usage with SystemCallFilter=
12850 in unit files and the new --system-call-filter= command line option
12851 of systemd-nspawn (see above).
12853 * ExecStart= lines in unit files gained two new modifiers: when a
12854 command line is prefixed with "!" the command will be executed as
12855 configured, except for the credentials applied by
12856 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
12857 "+", but does still apply namespacing options unlike "+". There's
12858 also "!!" now, which is mostly identical, but becomes a NOP on
12859 systems that support ambient capabilities. This is useful to write
12860 unit files that work with ambient capabilities where possible but
12861 automatically fall back to traditional privilege dropping mechanisms
12862 on systems where this is not supported.
12864 * ListenNetlink= settings in socket units now support RDMA netlink
12867 * A new unit file setting LockPersonality= has been added which permits
12868 locking down the chosen execution domain ("personality") of a service
12871 * A new special target "getty-pre.target" has been added, which is
12872 ordered before all text logins, and may be used to order services
12873 before textual logins acquire access to the console.
12875 * systemd will now attempt to load the virtio-rng.ko kernel module very
12876 early on if a VM environment supporting this is detected. This should
12877 improve entropy during early boot in virtualized environments.
12879 * A _netdev option is now supported in /etc/crypttab that operates in a
12880 similar way as the same option in /etc/fstab: it permits configuring
12881 encrypted devices that need to be ordered after the network is up.
12882 Following this logic, two new special targets
12883 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
12884 added that are to cryptsetup.target what remote-fs.target and
12885 remote-fs-pre.target are to local-fs.target.
12887 * Service units gained a new UnsetEnvironment= setting which permits
12888 unsetting specific environment variables for services that are
12889 normally passed to it (for example in order to mask out locale
12890 settings for specific services that can't deal with it).
12892 * Units acquired a new boolean option IPAccounting=. When turned on, IP
12893 traffic accounting (packet count as well as byte count) is done for
12894 the service, and shown as part of "systemctl status" or "systemd-run
12897 * Service units acquired two new options IPAddressAllow= and
12898 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
12899 for configuring a simple IP access control list for all sockets of
12900 the unit. These options are available also on .slice and .socket
12901 units, permitting flexible access list configuration for individual
12902 services as well as groups of services (as defined by a slice unit),
12903 including system-wide. Note that IP ACLs configured this way are
12904 enforced on every single IPv4 and IPv6 socket created by any process
12905 of the service unit, and apply to ingress as well as egress traffic.
12907 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
12908 structured log message is generated each time the unit is stopped,
12909 containing information about the consumed resources of this
12912 * A new setting KeyringMode= has been added to unit files, which may be
12913 used to control how the kernel keyring is set up for executed
12916 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
12917 "systemctl kexec" and "systemctl exit" are now always asynchronous in
12918 behaviour (that is: these commands return immediately after the
12919 operation was enqueued instead of waiting for the operation to
12920 complete). Previously, "systemctl poweroff" and "systemctl reboot"
12921 were asynchronous on systems using systemd-logind (i.e. almost
12922 always, and like they were on sysvinit), and the other three commands
12923 were unconditionally synchronous. With this release this is cleaned
12924 up, and callers will see the same asynchronous behaviour on all
12925 systems for all five operations.
12927 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
12930 * .timer units now accept calendar specifications in other timezones
12931 than UTC or the local timezone.
12933 * The tmpfiles snippet var.conf has been changed to create
12934 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
12935 the "utmp" group already, and it appears to be generally understood
12936 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
12937 databases. Previously this was implemented correctly for all these
12938 databases excepts btmp, which has been opened up like this now
12939 too. Note that while the other databases are world-readable
12940 (i.e. 0644), btmp is not and remains more restrictive.
12942 * The systemd-resolve tool gained a new --reset-server-features
12943 switch. When invoked like this systemd-resolved will forget
12944 everything it learnt about the features supported by the configured
12945 upstream DNS servers, and restarts the feature probing logic on the
12946 next resolver look-up for them at the highest feature level
12949 * The status dump systemd-resolved sends to the logs upon receiving
12950 SIGUSR1 now also includes information about all DNS servers it is
12951 configured to use, and the features levels it probed for them.
12953 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
12954 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
12955 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
12956 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
12957 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
12958 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
12959 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
12960 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
12961 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
12962 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
12963 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
12964 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
12965 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
12966 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
12967 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
12968 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
12969 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
12970 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
12971 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
12972 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
12974 — Berlin, 2017-10-06
12978 * Meson is now supported as build system in addition to Automake. It is
12979 our plan to remove Automake in one of our next releases, so that
12980 Meson becomes our exclusive build system. Hence, please start using
12981 the Meson build system in your downstream packaging. There's plenty
12982 of documentation around how to use Meson, the extremely brief
12985 ./autogen.sh && ./configure && make && sudo make install
12989 meson build && ninja -C build && sudo ninja -C build install
12991 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
12992 which permits configuring a timeout on the time a job is
12993 running. This is particularly useful for setting timeouts on jobs for
12996 * Unit files gained two new options ConditionUser= and ConditionGroup=
12997 for conditionalizing units based on the identity of the user/group
12998 running a systemd user instance.
13000 * systemd-networkd now understands a new FlowLabel= setting in the
13001 [VXLAN] section of .network files, as well as a Priority= in
13002 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
13003 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
13004 gained support for configuration of GENEVE links, and IPv6 address
13005 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
13007 * .link files now understand a new Port= setting.
13009 * systemd-networkd's DHCP support gained support for DHCP option 119
13010 (domain search list).
13012 * systemd-networkd gained support for serving IPv6 address ranges using
13013 the Router Advertisement protocol. The new .network configuration
13014 section [IPv6Prefix] may be used to configure the ranges to
13015 serve. This is implemented based on a new, minimal, native server
13016 implementation of RA.
13018 * journalctl's --output= switch gained support for a new parameter
13019 "short-iso-precise" for a mode where timestamps are shown as precise
13022 * systemd-udevd's "net_id" builtin may now generate stable network
13023 interface names from IBM PowerVM VIO devices as well as ACPI platform
13026 * MulticastDNS support in systemd-resolved may now be explicitly
13027 enabled/disabled using the new MulticastDNS= configuration file
13030 * systemd-resolved may now optionally use libidn2 instead of the libidn
13031 for processing internationalized domain names. Support for libidn2
13032 should be considered experimental and should not be enabled by
13035 * "machinectl pull-tar" and related call may now do verification of
13036 downloaded images using SUSE-style .sha256 checksum files in addition
13037 to the already existing support for validating using Ubuntu-style
13040 * sd-bus gained support for a new sd_bus_message_appendv() call which
13041 is va_list equivalent of sd_bus_message_append().
13043 * sd-boot gained support for validating images using SHIM/MOK.
13045 * The SMACK code learnt support for "onlycap".
13047 * systemd-mount --umount is now much smarter in figuring out how to
13048 properly unmount a device given its mount or device path.
13050 * The code to call libnss_dns as a fallback from libnss_resolve when
13051 the communication with systemd-resolved fails was removed. This
13052 fallback was redundant and interfered with the [!UNAVAIL=return]
13053 suffix. See nss-resolve(8) for the recommended configuration.
13055 * systemd-logind may now be restarted without losing state. It stores
13056 the file descriptors for devices it manages in the system manager
13057 using the FDSTORE= mechanism. Please note that further changes in
13058 other components may be required to make use of this (for example
13059 Xorg has code to listen for stops of systemd-logind and terminate
13060 itself when logind is stopped or restarted, in order to avoid using
13061 stale file descriptors for graphical devices, which is now
13062 counterproductive and must be reverted in order for restarts of
13063 systemd-logind to be safe. See
13064 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
13066 * All kernel-install plugins are called with the environment variable
13067 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
13068 /etc/machine-id. If the machine ID could not be determined,
13069 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
13070 anything in the entry directory (passed as the second argument) if
13071 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
13072 temporary directory is passed as the entry directory and removed
13073 after all the plugins exit.
13075 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
13076 will now use its value as the machine ID instead of the machine ID
13077 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
13078 /etc/machine-info and no machine ID is set in /etc/machine-id,
13079 kernel-install will try to store the current machine ID there as
13080 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
13081 will generate a new UUID, store it in /etc/machine-info as
13082 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
13084 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
13085 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
13086 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
13087 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
13088 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
13089 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
13090 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
13091 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
13092 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
13093 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
13094 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
13095 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
13096 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
13097 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
13098 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
13099 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
13100 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
13101 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
13102 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
13103 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
13104 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
13105 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
13106 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
13107 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
13108 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
13109 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
13110 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
13111 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
13112 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
13115 — Berlin, 2017-07-12
13119 * The "hybrid" control group mode has been modified to improve
13120 compatibility with "legacy" cgroups-v1 setups. Specifically, the
13121 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
13122 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
13123 cgroups-v1 hierarchy), the only externally visible change being that
13124 the cgroups-v2 hierarchy is also mounted, to
13125 /sys/fs/cgroup/unified. This should provide a large degree of
13126 compatibility with "legacy" cgroups-v1, while taking benefit of the
13127 better management capabilities of cgroups-v2.
13129 * The default control group setup mode may be selected both a boot-time
13130 via a set of kernel command line parameters (specifically:
13131 systemd.unified_cgroup_hierarchy= and
13132 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
13133 default selected on the configure command line
13134 (--with-default-hierarchy=). The upstream default is "hybrid"
13135 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
13136 this will change in a future systemd version to be "unified" (pure
13137 cgroups-v2 mode). The third option for the compile time option is
13138 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
13139 distributions to default to "hybrid" mode for release distributions,
13140 starting with v233. We recommend "unified" for development
13141 distributions (specifically: distributions such as Fedora's rawhide)
13142 as that's where things are headed in the long run. Use "legacy" for
13143 greatest stability and compatibility only.
13145 * Note one current limitation of "unified" and "hybrid" control group
13146 setup modes: the kernel currently does not permit the systemd --user
13147 instance (i.e. unprivileged code) to migrate processes between two
13148 disconnected cgroup subtrees, even if both are managed and owned by
13149 the user. This effectively means "systemd-run --user --scope" doesn't
13150 work when invoked from outside of any "systemd --user" service or
13151 scope. Specifically, it is not supported from session scopes. We are
13152 working on fixing this in a future systemd version. (See #3388 for
13153 further details about this.)
13155 * DBus policy files are now installed into /usr rather than /etc. Make
13156 sure your system has dbus >= 1.9.18 running before upgrading to this
13157 version, or override the install path with --with-dbuspolicydir= .
13159 * All python scripts shipped with systemd (specifically: the various
13160 tests written in Python) now require Python 3.
13162 * systemd unit tests can now run standalone (without the source or
13163 build directories), and can be installed into /usr/lib/systemd/tests/
13164 with 'make install-tests'.
13166 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
13167 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
13170 * Support for the %c, %r, %R specifiers in unit files has been
13171 removed. Specifiers are not supposed to be dependent on configuration
13172 in the unit file itself (so that they resolve the same regardless
13173 where used in the unit files), but these specifiers were influenced
13174 by the Slice= option.
13176 * The shell invoked by debug-shell.service now defaults to /bin/sh in
13177 all cases. If distributions want to use a different shell for this
13178 purpose (for example Fedora's /sbin/sushell) they need to specify
13179 this explicitly at configure time using --with-debug-shell=.
13181 * The confirmation spawn prompt has been reworked to offer the
13184 (c)ontinue, proceed without asking anymore
13185 (D)ump, show the state of the unit
13186 (f)ail, don't execute the command and pretend it failed
13188 (i)nfo, show a short summary of the unit
13189 (j)obs, show jobs that are in progress
13190 (s)kip, don't execute the command and pretend it succeeded
13191 (y)es, execute the command
13193 The 'n' choice for the confirmation spawn prompt has been removed,
13194 because its meaning was confusing.
13196 The prompt may now also be redirected to an alternative console by
13197 specifying the console as parameter to systemd.confirm_spawn=.
13199 * Services of Type=notify require a READY=1 notification to be sent
13200 during startup. If no such message is sent, the service now fails,
13201 even if the main process exited with a successful exit code.
13203 * Services that fail to start up correctly now always have their
13204 ExecStopPost= commands executed. Previously, they'd enter "failed"
13205 state directly, without executing these commands.
13207 * The option MulticastDNS= of network configuration files has acquired
13208 an actual implementation. With MulticastDNS=yes a host can resolve
13209 names of remote hosts and reply to mDNS A and AAAA requests.
13211 * When units are about to be started an additional check is now done to
13212 ensure that all dependencies of type BindsTo= (when used in
13213 combination with After=) have been started.
13215 * systemd-analyze gained a new verb "syscall-filter" which shows which
13216 system call groups are defined for the SystemCallFilter= unit file
13217 setting, and which system calls they contain.
13219 * A new system call filter group "@filesystem" has been added,
13220 consisting of various file system related system calls. Group
13221 "@reboot" has been added, covering reboot, kexec and shutdown related
13222 calls. Finally, group "@swap" has been added covering swap
13223 configuration related calls.
13225 * A new unit file option RestrictNamespaces= has been added that may be
13226 used to restrict access to the various process namespace types the
13227 Linux kernel provides. Specifically, it may be used to take away the
13228 right for a service unit to create additional file system, network,
13229 user, and other namespaces. This sandboxing option is particularly
13230 relevant due to the high amount of recently discovered namespacing
13231 related vulnerabilities in the kernel.
13233 * systemd-udev's .link files gained support for a new AutoNegotiation=
13234 setting for configuring Ethernet auto-negotiation.
13236 * systemd-networkd's .network files gained support for a new
13237 ListenPort= setting in the [DHCP] section to explicitly configure the
13238 UDP client port the DHCP client shall listen on.
13240 * .network files gained a new Unmanaged= boolean setting for explicitly
13241 excluding one or more interfaces from management by systemd-networkd.
13243 * The systemd-networkd ProxyARP= option has been renamed to
13244 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
13245 renamed to ReduceARPProxy=. The old names continue to be available
13248 * systemd-networkd gained support for configuring IPv6 Proxy NDP
13249 addresses via the new IPv6ProxyNDPAddress= .network file setting.
13251 * systemd-networkd's bonding device support gained support for two new
13252 configuration options ActiveSlave= and PrimarySlave=.
13254 * The various options in the [Match] section of .network files gained
13255 support for negative matching.
13257 * New systemd-specific mount options are now understood in /etc/fstab:
13259 x-systemd.mount-timeout= may be used to configure the maximum
13260 permitted runtime of the mount command.
13262 x-systemd.device-bound may be set to bind a mount point to its
13263 backing device unit, in order to automatically remove a mount point
13264 if its backing device is unplugged. This option may also be
13265 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
13266 on the block device, which is now automatically set for all CDROM
13267 drives, so that mounted CDs are automatically unmounted when they are
13268 removed from the drive.
13270 x-systemd.after= and x-systemd.before= may be used to explicitly
13271 order a mount after or before another unit or mount point.
13273 * Enqueued start jobs for device units are now automatically garbage
13274 collected if there are no jobs waiting for them anymore.
13276 * systemctl list-jobs gained two new switches: with --after, for every
13277 queued job the jobs it's waiting for are shown; with --before the
13278 jobs which it's blocking are shown.
13280 * systemd-nspawn gained support for ephemeral boots from disk images
13281 (or in other words: --ephemeral and --image= may now be
13282 combined). Moreover, ephemeral boots are now supported for normal
13283 directories, even if the backing file system is not btrfs. Of course,
13284 if the file system does not support file system snapshots or
13285 reflinks, the initial copy operation will be relatively expensive, but
13286 this should still be suitable for many use cases.
13288 * Calendar time specifications in .timer units now support
13289 specifications relative to the end of a month by using "~" instead of
13290 "-" as separator between month and day. For example, "*-02~03" means
13291 "the third last day in February". In addition a new syntax for
13292 repeated events has been added using the "/" character. For example,
13293 "9..17/2:00" means "every two hours from 9am to 5pm".
13295 * systemd-socket-proxyd gained a new parameter --connections-max= for
13296 configuring the maximum number of concurrent connections.
13298 * sd-id128 gained a new API for generating unique IDs for the host in a
13299 way that does not leak the machine ID. Specifically,
13300 sd_id128_get_machine_app_specific() derives an ID based on the
13301 machine ID in a well-defined, non-reversible, stable way. This is
13302 useful whenever an identifier for the host is needed but where the
13303 identifier shall not be useful to identify the system beyond the
13304 scope of the application itself. (Internally this uses HMAC-SHA256 as
13305 keyed hash function using the machine ID as input.)
13307 * NotifyAccess= gained a new supported value "exec". When set
13308 notifications are accepted from all processes systemd itself invoked,
13309 including all control processes.
13311 * .nspawn files gained support for defining overlay mounts using the
13312 Overlay= and OverlayReadOnly= options. Previously this functionality
13313 was only available on the systemd-nspawn command line.
13315 * systemd-nspawn's --bind= and --overlay= options gained support for
13316 bind/overlay mounts whose source lies within the container tree by
13317 prefixing the source path with "+".
13319 * systemd-nspawn's --bind= and --overlay= options gained support for
13320 automatically allocating a temporary source directory in /var/tmp
13321 that is removed when the container dies. Specifically, if the source
13322 directory is specified as empty string this mechanism is selected. An
13323 example usage is --overlay=+/var::/var, which creates an overlay
13324 mount based on the original /var contained in the image, overlaid
13325 with a temporary directory in the host's /var/tmp. This way changes
13326 to /var are automatically flushed when the container shuts down.
13328 * systemd-nspawn --image= option does now permit raw file system block
13329 devices (in addition to images containing partition tables, as
13332 * The disk image dissection logic in systemd-nspawn gained support for
13333 automatically setting up LUKS encrypted as well as Verity protected
13334 partitions. When a container is booted from an encrypted image the
13335 passphrase is queried at start-up time. When a container with Verity
13336 data is started, the root hash is search in a ".roothash" file
13337 accompanying the disk image (alternatively, pass the root hash via
13338 the new --root-hash= command line option).
13340 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
13341 be used to dissect disk images the same way as systemd-nspawn does
13342 it, following the Bootable Partition Specification. It may even be
13343 used to mount disk images with complex partition setups (including
13344 LUKS and Verity partitions) to a local host directory, in order to
13345 inspect them. This tool is not considered public API (yet), and is
13346 thus not installed into /usr/bin. Please do not rely on its
13347 existence, since it might go away or be changed in later systemd
13350 * A new generator "systemd-verity-generator" has been added, similar in
13351 style to "systemd-cryptsetup-generator", permitting automatic setup of
13352 Verity root partitions when systemd boots up. In order to make use of
13353 this your partition setup should follow the Discoverable Partitions
13354 Specification, and the GPT partition ID of the root file system
13355 partition should be identical to the upper 128-bit of the Verity root
13356 hash. The GPT partition ID of the Verity partition protecting it
13357 should be the lower 128-bit of the Verity root hash. If the partition
13358 image follows this model it is sufficient to specify a single
13359 "roothash=" kernel command line argument to both configure which root
13360 image and verity partition to use as well as the root hash for
13361 it. Note that systemd-nspawn's Verity support follows the same
13362 semantics, meaning that disk images with proper Verity data in place
13363 may be booted in containers with systemd-nspawn as well as on
13364 physical systems via the verity generator. Also note that the "mkosi"
13365 tool available at https://github.com/systemd/mkosi has been updated
13366 to generate Verity protected disk images following this scheme. In
13367 fact, it has been updated to generate disk images that optionally
13368 implement a complete UEFI SecureBoot trust chain, involving a signed
13369 kernel and initrd image that incorporates such a root hash as well as
13370 a Verity-enabled root partition.
13372 * The hardware database (hwdb) udev supports has been updated to carry
13373 accelerometer quirks.
13375 * All system services are now run with a fresh kernel keyring set up
13376 for them. The invocation ID is stored by default in it, thus
13377 providing a safe, non-overridable way to determine the invocation
13378 ID of each service.
13380 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
13381 options for bind mounting arbitrary paths in a service-specific
13382 way. When these options are used, arbitrary host or service files and
13383 directories may be mounted to arbitrary locations in the service's
13386 * Documentation has been added that lists all of systemd's low-level
13387 environment variables:
13389 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
13391 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
13392 whether a specific socket file descriptor matches a specified socket
13395 * systemd-firstboot has been updated to check for the
13396 systemd.firstboot= kernel command line option. It accepts a boolean
13397 and when set to false the first boot questions are skipped.
13399 * systemd-fstab-generator has been updated to check for the
13400 systemd.volatile= kernel command line option, which either takes an
13401 optional boolean parameter or the special value "state". If used the
13402 system may be booted in a "volatile" boot mode. Specifically,
13403 "systemd.volatile" is used, the root directory will be mounted as
13404 tmpfs, and only /usr is mounted from the actual root file system. If
13405 "systemd.volatile=state" is used, the root directory will be mounted
13406 as usual, but /var is mounted as tmpfs. This concept provides similar
13407 functionality as systemd-nspawn's --volatile= option, but provides it
13408 on physical boots. Use this option for implementing stateless
13409 systems, or testing systems with all state and/or configuration reset
13410 to the defaults. (Note though that many distributions are not
13411 prepared to boot up without a populated /etc or /var, though.)
13413 * systemd-gpt-auto-generator gained support for LUKS encrypted root
13414 partitions. Previously it only supported LUKS encrypted partitions
13415 for all other uses, except for the root partition itself.
13417 * Socket units gained support for listening on AF_VSOCK sockets for
13418 communication in virtualized QEMU environments.
13420 * The "configure" script gained a new option --with-fallback-hostname=
13421 for specifying the fallback hostname to use if none is configured in
13422 /etc/hostname. For example, by specifying
13423 --with-fallback-hostname=fedora it is possible to default to a
13424 hostname of "fedora" on pristine installations.
13426 * systemd-cgls gained support for a new --unit= switch for listing only
13427 the control groups of a specific unit. Similar --user-unit= has been
13428 added for listing only the control groups of a specific user unit.
13430 * systemd-mount gained a new --umount switch for unmounting a mount or
13431 automount point (and all mount/automount points below it).
13433 * systemd will now refuse full configuration reloads (via systemctl
13434 daemon-reload and related calls) unless at least 16MiB of free space
13435 are available in /run. This is a safety precaution in order to ensure
13436 that generators can safely operate after the reload completed.
13438 * A new unit file option RootImage= has been added, which has a similar
13439 effect as RootDirectory= but mounts the service's root directory from
13440 a disk image instead of plain directory. This logic reuses the same
13441 image dissection and mount logic that systemd-nspawn already uses,
13442 and hence supports any disk images systemd-nspawn supports, including
13443 those following the Discoverable Partition Specification, as well as
13444 Verity enabled images. This option enables systemd to run system
13445 services directly off disk images acting as resource bundles,
13446 possibly even including full integrity data.
13448 * A new MountAPIVFS= unit file option has been added, taking a boolean
13449 argument. If enabled /proc, /sys and /dev (collectively called the
13450 "API VFS") will be mounted for the service. This is only relevant if
13451 RootDirectory= or RootImage= is used for the service, as these mounts
13452 are of course in place in the host mount namespace anyway.
13454 * systemd-nspawn gained support for a new --pivot-root= switch. If
13455 specified the root directory within the container image is pivoted to
13456 the specified mount point, while the original root disk is moved to a
13457 different place. This option enables booting of ostree images
13458 directly with systemd-nspawn.
13460 * The systemd build scripts will no longer complain if the NTP server
13461 addresses are not changed from the defaults. Google now supports
13462 these NTP servers officially. We still recommend downstreams to
13463 properly register an NTP pool with the NTP pool project though.
13465 * coredumpctl gained a new "--reverse" option for printing the list
13466 of coredumps in reverse order.
13468 * coredumpctl will now show additional information about truncated and
13469 inaccessible coredumps, as well as coredumps that are still being
13470 processed. It also gained a new --quiet switch for suppressing
13471 additional informational message in its output.
13473 * coredumpctl gained support for only showing coredumps newer and/or
13474 older than specific timestamps, using the new --since= and --until=
13475 options, reminiscent of journalctl's options by the same name.
13477 * The systemd-coredump logic has been improved so that it may be reused
13478 to collect backtraces in non-compiled languages, for example in
13479 scripting languages such as Python.
13481 * machinectl will now show the UID shift of local containers, if user
13482 namespacing is enabled for them.
13484 * systemd will now optionally run "environment generator" binaries at
13485 configuration load time. They may be used to add environment
13486 variables to the environment block passed to services invoked. One
13487 user environment generator is shipped by default that sets up
13488 environment variables based on files dropped into /etc/environment.d
13489 and ~/.config/environment.d/.
13491 * systemd-resolved now includes the new, recently published 2017 DNSSEC
13494 * hostnamed has been updated to report a new chassis type of
13495 "convertible" to cover "foldable" laptops that can both act as a
13496 tablet and as a laptop, such as various Lenovo Yoga devices.
13498 Contributions from: Adrián López, Alexander Galanin, Alexander
13499 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
13500 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
13501 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
13502 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
13503 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
13504 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
13505 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
13506 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
13507 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
13508 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
13509 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
13510 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
13511 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
13512 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
13513 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
13514 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
13515 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
13516 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
13517 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
13518 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
13519 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
13520 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
13521 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
13522 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
13523 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
13524 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
13525 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
13528 — Berlin, 2017-03-01
13532 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
13533 RestrictAddressFamilies= enabled. These sandboxing options should
13534 generally be compatible with the various external udev call-out
13535 binaries we are aware of, however there may be exceptions, in
13536 particular when exotic languages for these call-outs are used. In
13537 this case, consider turning off these settings locally.
13539 * The new RemoveIPC= option can be used to remove IPC objects owned by
13540 the user or group of a service when that service exits.
13542 * The new ProtectKernelModules= option can be used to disable explicit
13543 load and unload operations of kernel modules by a service. In
13544 addition access to /usr/lib/modules is removed if this option is set.
13546 * ProtectSystem= option gained a new value "strict", which causes the
13547 whole file system tree with the exception of /dev, /proc, and /sys,
13548 to be remounted read-only for a service.
13550 * The new ProtectKernelTunables= option can be used to disable
13551 modification of configuration files in /sys and /proc by a service.
13552 Various directories and files are remounted read-only, so access is
13553 restricted even if the file permissions would allow it.
13555 * The new ProtectControlGroups= option can be used to disable write
13556 access by a service to /sys/fs/cgroup.
13558 * Various systemd services have been hardened with
13559 ProtectKernelTunables=yes, ProtectControlGroups=yes,
13560 RestrictAddressFamilies=.
13562 * Support for dynamically creating users for the lifetime of a service
13563 has been added. If DynamicUser=yes is specified, user and group IDs
13564 will be allocated from the range 61184…65519 for the lifetime of the
13565 service. They can be resolved using the new nss-systemd.so NSS
13566 module. The module must be enabled in /etc/nsswitch.conf. Services
13567 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
13568 any resources allocated by the service will be cleaned up when the
13569 service exits. They also have ProtectHome=read-only and
13570 ProtectSystem=strict enabled, so they are not able to make any
13571 permanent modifications to the system.
13573 * The nss-systemd module also always resolves root and nobody, making
13574 it possible to have no /etc/passwd or /etc/group files in minimal
13575 container or chroot environments.
13577 * Services may be started with their own user namespace using the new
13578 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
13579 under which the service is running are mapped. All other users are
13582 * Support for the cgroup namespace has been added to systemd-nspawn. If
13583 supported by kernel, the container system started by systemd-nspawn
13584 will have its own view of the cgroup hierarchy. This new behaviour
13585 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
13587 * The new MemorySwapMax= option can be used to limit the maximum swap
13588 usage under the unified cgroup hierarchy.
13590 * Support for the CPU controller in the unified cgroup hierarchy has
13591 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
13592 options. This controller requires out-of-tree patches for the kernel
13593 and the support is provisional.
13595 * Mount and automount units may now be created transiently
13596 (i.e. dynamically at runtime via the bus API, instead of requiring
13597 unit files in the file system).
13599 * systemd-mount is a new tool which may mount file systems – much like
13600 mount(8), optionally pulling in additional dependencies through
13601 transient .mount and .automount units. For example, this tool
13602 automatically runs fsck on a backing block device before mounting,
13603 and allows the automount logic to be used dynamically from the
13604 command line for establishing mount points. This tool is particularly
13605 useful when dealing with removable media, as it will ensure fsck is
13606 run – if necessary – before the first access and that the file system
13607 is quickly unmounted after each access by utilizing the automount
13608 logic. This maximizes the chance that the file system on the
13609 removable media stays in a clean state, and if it isn't in a clean
13610 state is fixed automatically.
13612 * LazyUnmount=yes option for mount units has been added to expose the
13613 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
13616 * /efi will be used as the mount point of the EFI boot partition, if
13617 the directory is present, and the mount point was not configured
13618 through other means (e.g. fstab). If /efi directory does not exist,
13619 /boot will be used as before. This makes it easier to automatically
13620 mount the EFI partition on systems where /boot is used for something
13623 * When operating on GPT disk images for containers, systemd-nspawn will
13624 now mount the ESP to /boot or /efi according to the same rules as PID
13625 1 running on a host. This allows tools like "bootctl" to operate
13626 correctly within such containers, in order to make container images
13627 bootable on physical systems.
13629 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
13631 * Two new user session targets have been added to support running
13632 graphical sessions under the systemd --user instance:
13633 graphical-session.target and graphical-session-pre.target. See
13634 systemd.special(7) for a description of how those targets should be
13637 * The vconsole initialization code has been significantly reworked to
13638 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
13639 support unicode keymaps. Font and keymap configuration will now be
13640 copied to all allocated virtual consoles.
13642 * FreeBSD's bhyve virtualization is now detected.
13644 * Information recorded in the journal for core dumps now includes the
13645 contents of /proc/mountinfo and the command line of the process at
13646 the top of the process hierarchy (which is usually the init process
13649 * systemd-journal-gatewayd learned the --directory= option to serve
13650 files from the specified location.
13652 * journalctl --root=… can be used to peruse the journal in the
13653 /var/log/ directories inside of a container tree. This is similar to
13654 the existing --machine= option, but does not require the container to
13657 * The hardware database has been extended to support
13658 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
13661 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
13662 specify the click rate for mice which include a horizontal wheel with
13663 a click rate that is different than the one for the vertical wheel.
13665 * systemd-run gained a new --wait option that makes service execution
13666 synchronous. (Specifically, the command will not return until the
13667 specified service binary exited.)
13669 * systemctl gained a new --wait option that causes the start command to
13670 wait until the units being started have terminated again.
13672 * A new journal output mode "short-full" has been added which displays
13673 timestamps with abbreviated English day names and adds a timezone
13674 suffix. Those timestamps include more information than the default
13675 "short" output mode, and can be passed directly to journalctl's
13676 --since= and --until= options.
13678 * /etc/resolv.conf will be bind-mounted into containers started by
13679 systemd-nspawn, if possible, so any changes to resolv.conf contents
13680 are automatically propagated to the container.
13682 * The number of instances for socket-activated services originating
13683 from a single IP address can be limited with
13684 MaxConnectionsPerSource=, extending the existing setting of
13687 * systemd-networkd gained support for vcan ("Virtual CAN") interface
13690 * .netdev and .network configuration can now be extended through
13693 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
13694 Segmentation Offload, Generic Receive Offload, Large Receive Offload
13695 can be enabled and disabled using the new UDPSegmentationOffload=,
13696 TCPSegmentationOffload=, GenericSegmentationOffload=,
13697 GenericReceiveOffload=, LargeReceiveOffload= options in the
13698 [Link] section of .link files.
13700 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
13701 Port VLAN ID can be configured for bridge devices using the new STP=,
13702 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
13703 section of .netdev files.
13705 * The route table to which routes received over DHCP or RA should be
13706 added can be configured with the new RouteTable= option in the [DHCP]
13707 and [IPv6AcceptRA] sections of .network files.
13709 * The Address Resolution Protocol can be disabled on links managed by
13710 systemd-networkd using the ARP=no setting in the [Link] section of
13713 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
13714 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
13715 encode information about the result and exit codes of the current
13716 service runtime cycle.
13718 * systemd-sysctl will now configure kernel parameters in the order
13719 they occur in the configuration files. This matches what sysctl
13720 has been traditionally doing.
13722 * kernel-install "plugins" that are executed to perform various
13723 tasks after a new kernel is added and before an old one is removed
13724 can now return a special value to terminate the procedure and
13725 prevent any later plugins from running.
13727 * Journald's SplitMode=login setting has been deprecated. It has been
13728 removed from documentation, and its use is discouraged. In a future
13729 release it will be completely removed, and made equivalent to current
13730 default of SplitMode=uid.
13732 * Storage=both option setting in /etc/systemd/coredump.conf has been
13733 removed. With fast LZ4 compression storing the core dump twice is not
13736 * The --share-system systemd-nspawn option has been replaced with an
13737 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
13738 this functionality is discouraged. In addition the variables
13739 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
13740 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
13741 individual namespaces.
13743 * "machinectl list" now shows the IP address of running containers in
13744 the output, as well as OS release information.
13746 * "loginctl list" now shows the TTY of each session in the output.
13748 * sd-bus gained new API calls sd_bus_track_set_recursive(),
13749 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
13750 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
13751 tracking objects in a "recursive" mode, where a single client can be
13752 counted multiple times, if it takes multiple references.
13754 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
13755 sd_bus_get_exit_on_disconnect(). They may be used to make a
13756 process using sd-bus automatically exit if the bus connection is
13759 * Bus clients of the service manager may now "pin" loaded units into
13760 memory, by taking an explicit reference on them. This is useful to
13761 ensure the client can retrieve runtime data about the service even
13762 after the service completed execution. Taking such a reference is
13763 available only for privileged clients and should be helpful to watch
13764 running services in a race-free manner, and in particular collect
13765 information about exit statuses and results.
13767 * The nss-resolve module has been changed to strictly return UNAVAIL
13768 when communication via D-Bus with resolved failed, and NOTFOUND when
13769 a lookup completed but was negative. This means it is now possible to
13770 neatly configure fallbacks using nsswitch.conf result checking
13771 expressions. Taking benefit of this, the new recommended
13772 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
13774 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
13776 * A new setting CtrlAltDelBurstAction= has been added to
13777 /etc/systemd/system.conf which may be used to configure the precise
13778 behaviour if the user on the console presses Ctrl-Alt-Del more often
13779 than 7 times in 2s. Previously this would unconditionally result in
13780 an expedited, immediate reboot. With this new setting the precise
13781 operation may be configured in more detail, and also turned off
13784 * In .netdev files two new settings RemoteChecksumTx= and
13785 RemoteChecksumRx= are now understood that permit configuring the
13786 remote checksumming logic for VXLAN networks.
13788 * The service manager learnt a new "invocation ID" concept for invoked
13789 services. Each runtime cycle of a service will get a new invocation
13790 ID (a 128-bit random UUID) assigned that identifies the current
13791 run of the service uniquely and globally. A new invocation ID
13792 is generated each time a service starts up. The journal will store
13793 the invocation ID of a service along with any logged messages, thus
13794 making the invocation ID useful for matching the online runtime of a
13795 service with the offline log data it generated in a safe way without
13796 relying on synchronized timestamps. In many ways this new service
13797 invocation ID concept is similar to the kernel's boot ID concept that
13798 uniquely and globally identifies the runtime of each boot. The
13799 invocation ID of a service is passed to the service itself via an
13800 environment variable ($INVOCATION_ID). A new bus call
13801 GetUnitByInvocationID() has been added that is similar to GetUnit()
13802 but instead of retrieving the bus path for a unit by its name
13803 retrieves it by its invocation ID. The returned path is valid only as
13804 long as the passed invocation ID is current.
13806 * systemd-resolved gained a new "DNSStubListener" setting in
13807 resolved.conf. It either takes a boolean value or the special values
13808 "udp" and "tcp", and configures whether to enable the stub DNS
13809 listener on 127.0.0.53:53.
13811 * IP addresses configured via networkd may now carry additional
13812 configuration settings supported by the kernel. New options include:
13813 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
13814 PrefixRoute=, AutoJoin=.
13816 * The PAM configuration fragment file for "user@.service" shipped with
13817 systemd (i.e. the --user instance of systemd) has been stripped to
13818 the minimum necessary to make the system boot. Previously, it
13819 contained Fedora-specific stanzas that did not apply to other
13820 distributions. It is expected that downstream distributions add
13821 additional configuration lines, matching their needs to this file,
13822 using it only as rough template of what systemd itself needs. Note
13823 that this reduced fragment does not even include an invocation of
13824 pam_limits which most distributions probably want to add, even though
13825 systemd itself does not need it. (There's also the new build time
13826 option --with-pamconfdir=no to disable installation of the PAM
13827 fragment entirely.)
13829 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
13830 capability is now also dropped from its set (in addition to
13831 CAP_SYS_MKNOD as before).
13833 * In service unit files it is now possible to connect a specific named
13834 file descriptor with stdin/stdout/stdout of an executed service. The
13835 name may be specified in matching .socket units using the
13836 FileDescriptorName= setting.
13838 * A number of journal settings may now be configured on the kernel
13839 command line. Specifically, the following options are now understood:
13840 systemd.journald.max_level_console=,
13841 systemd.journald.max_level_store=,
13842 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
13843 systemd.journald.max_level_wall=.
13845 * "systemctl is-enabled --full" will now show by which symlinks a unit
13846 file is enabled in the unit dependency tree.
13848 * Support for VeraCrypt encrypted partitions has been added to the
13849 "cryptsetup" logic and /etc/crypttab.
13851 * systemd-detect-virt gained support for a new --private-users switch
13852 that checks whether the invoking processes are running inside a user
13853 namespace. Similar, a new special value "private-users" for the
13854 existing ConditionVirtualization= setting has been added, permitting
13855 skipping of specific units in user namespace environments.
13857 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
13858 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
13859 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
13860 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
13861 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
13862 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
13863 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
13864 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
13865 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
13866 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
13867 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
13868 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
13869 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
13870 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
13871 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
13872 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
13873 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
13874 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
13875 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
13876 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
13877 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
13878 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
13879 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
13880 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
13881 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
13882 Jędrzejewski-Szmek, Zeal Jagannatha
13884 — Santa Fe, 2016-11-03
13888 * In service units the various ExecXYZ= settings have been extended
13889 with an additional special character as first argument of the
13890 assigned value: if the character '+' is used the specified command
13891 line it will be run with full privileges, regardless of User=,
13892 Group=, CapabilityBoundingSet= and similar options. The effect is
13893 similar to the existing PermissionsStartOnly= option, but allows
13894 configuration of this concept for each executed command line
13897 * Services may now alter the service watchdog timeout at runtime by
13898 sending a WATCHDOG_USEC= message via sd_notify().
13900 * MemoryLimit= and related unit settings now optionally take percentage
13901 specifications. The percentage is taken relative to the amount of
13902 physical memory in the system (or in case of containers, the assigned
13903 amount of memory). This allows scaling service resources neatly with
13904 the amount of RAM available on the system. Similarly, systemd-logind's
13905 RuntimeDirectorySize= option now also optionally takes percentage
13908 * In similar fashion TasksMax= takes percentage values now, too. The
13909 value is taken relative to the configured maximum number of processes
13910 on the system. The per-service task maximum has been changed to 15%
13911 using this functionality. (Effectively this is an increase of 512 →
13912 4915 for service units, given the kernel's default pid_max setting.)
13914 * Calendar time specifications in .timer units now understand a ".."
13915 syntax for time ranges. Example: "4..7:10" may now be used for
13916 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
13919 * The InaccessableDirectories=, ReadOnlyDirectories= and
13920 ReadWriteDirectories= unit file settings have been renamed to
13921 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
13922 applied to all kinds of file nodes, and not just directories, with
13923 the exception of symlinks. Specifically these settings may now be
13924 used on block and character device nodes, UNIX sockets and FIFOS as
13925 well as regular files. The old names of these settings remain
13926 available for compatibility.
13928 * systemd will now log about all service processes it kills forcibly
13929 (using SIGKILL) because they remained after the clean shutdown phase
13930 of the service completed. This should help identifying services that
13931 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
13932 systemd-logind's configuration a similar log message is generated for
13933 processes killed at the end of each session due to this setting.
13935 * systemd will now set the $JOURNAL_STREAM environment variable for all
13936 services whose stdout/stderr are connected to the Journal (which
13937 effectively means by default: all services). The variable contains
13938 the device and inode number of the file descriptor used for
13939 stdout/stderr. This may be used by invoked programs to detect whether
13940 their stdout/stderr is connected to the Journal, in which case they
13941 can switch over to direct Journal communication, thus being able to
13942 pass extended, structured metadata along with their log messages. As
13943 one example, this is now used by glib's logging primitives.
13945 * When using systemd's default tmp.mount unit for /tmp, the mount point
13946 will now be established with the "nosuid" and "nodev" options. This
13947 avoids privilege escalation attacks that put traps and exploits into
13948 /tmp. However, this might cause problems if you e.g. put container
13949 images or overlays into /tmp; if you need this, override tmp.mount's
13950 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
13953 * systemd now supports the "memory" cgroup controller also on
13956 * The systemd-cgtop tool now optionally takes a control group path as
13957 command line argument. If specified, the control group list shown is
13958 limited to subgroups of that group.
13960 * The SystemCallFilter= unit file setting gained support for
13961 pre-defined, named system call filter sets. For example
13962 SystemCallFilter=@clock is now an effective way to make all clock
13963 changing-related system calls unavailable to a service. A number of
13964 similar pre-defined groups are defined. Writing system call filters
13965 for system services is simplified substantially with this new
13966 concept. Accordingly, all of systemd's own, long-running services now
13967 enable system call filtering based on this, by default.
13969 * A new service setting MemoryDenyWriteExecute= has been added, taking
13970 a boolean value. If turned on, a service may no longer create memory
13971 mappings that are writable and executable at the same time. This
13972 enhances security for services where this is enabled as it becomes
13973 harder to dynamically write and then execute memory in exploited
13974 service processes. This option has been enabled for all of systemd's
13975 own long-running services.
13977 * A new RestrictRealtime= service setting has been added, taking a
13978 boolean argument. If set the service's processes may no longer
13979 acquire realtime scheduling. This improves security as realtime
13980 scheduling may otherwise be used to easily freeze the system.
13982 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
13983 value. This may be used for requesting that the system manager inside
13984 of the container reports start-up completion to nspawn which then
13985 propagates this notification further to the service manager
13986 supervising nspawn itself. A related option NotifyReady= in .nspawn
13987 files has been added too. This functionality allows ordering of the
13988 start-up of multiple containers using the usual systemd ordering
13991 * machinectl gained a new command "stop" that is an alias for
13994 * systemd-resolved gained support for contacting DNS servers on
13995 link-local IPv6 addresses.
13997 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
13998 its caches. A method call for requesting the same operation has been
13999 added to the bus API too, and is made available via "systemd-resolve
14002 * systemd-resolve gained a new --status switch. If passed a brief
14003 summary of the used DNS configuration with per-interface information
14006 * resolved.conf gained a new Cache= boolean option, defaulting to
14007 on. If turned off local DNS caching is disabled. This comes with a
14008 performance penalty in particular when DNSSEC is enabled. Note that
14009 resolved disables its internal caching implicitly anyway, when the
14010 configured DNS server is on a host-local IP address such as ::1 or
14011 127.0.0.1, thus automatically avoiding double local caching.
14013 * systemd-resolved now listens on the local IP address 127.0.0.53:53
14014 for DNS requests. This improves compatibility with local programs
14015 that do not use the libc NSS or systemd-resolved's bus APIs for name
14016 resolution. This minimal DNS service is only available to local
14017 programs and does not implement the full DNS protocol, but enough to
14018 cover local DNS clients. A new, static resolv.conf file, listing just
14019 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
14020 now recommended to make /etc/resolv.conf a symlink to this file in
14021 order to route all DNS lookups to systemd-resolved, regardless if
14022 done via NSS, the bus API or raw DNS packets. Note that this local
14023 DNS service is not as fully featured as the libc NSS or
14024 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
14025 used to deliver link-local address information (as this implies
14026 sending a local interface index along), LLMNR/mDNS support via this
14027 interface is severely restricted. It is thus strongly recommended for
14028 all applications to use the libc NSS API or native systemd-resolved
14031 * systemd-networkd's bridge support learned a new setting
14032 VLANFiltering= for controlling VLAN filtering. Moreover a new section
14033 in .network files has been added for configuring VLAN bridging in
14034 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
14036 * systemd-networkd's IPv6 Router Advertisement code now makes use of
14037 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
14038 now be acquired without relying on DHCPv6. Two new options
14039 UseDomains= and UseDNS= have been added to configure this behaviour.
14041 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
14042 renamed IPv6AcceptRA=, without altering its behaviour. The old
14043 setting name remains available for compatibility reasons.
14045 * The systemd-networkd VTI/VTI6 tunneling support gained new options
14046 Key=, InputKey= and OutputKey=.
14048 * systemd-networkd gained support for VRF ("Virtual Routing Function")
14049 interface configuration.
14051 * "systemctl edit" may now be used to create new unit files by
14052 specifying the --force switch.
14054 * sd-event gained a new function sd_event_get_iteration() for
14055 requesting the current iteration counter of the event loop. It starts
14056 at zero and is increased by one with each event loop iteration.
14058 * A new rpm macro %systemd_ordering is provided by the macros.systemd
14059 file. It can be used in lieu of %systemd_requires in packages which
14060 don't use any systemd functionality and are intended to be installed
14061 in minimal containers without systemd present. This macro provides
14062 ordering dependencies to ensure that if the package is installed in
14063 the same rpm transaction as systemd, systemd will be installed before
14064 the scriptlets for the package are executed, allowing unit presets
14067 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
14068 been added to simplify packaging of generators.
14070 * The os-release file gained VERSION_CODENAME field for the
14071 distribution nickname (e.g. VERSION_CODENAME=woody).
14073 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
14074 can be set to disable parsing of metadata and the creation
14075 of persistent symlinks for that device.
14077 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
14078 to make them available to logged-in users has been reverted.
14080 * Much of the common code of the various systemd components is now
14081 built into an internal shared library libsystemd-shared-231.so
14082 (incorporating the systemd version number in the name, to be updated
14083 with future releases) that the components link to. This should
14084 decrease systemd footprint both in memory during runtime and on
14085 disk. Note that the shared library is not for public use, and is
14086 neither API nor ABI stable, but is likely to change with every new
14087 released update. Packagers need to make sure that binaries
14088 linking to libsystemd-shared.so are updated in step with the
14091 * Configuration for "mkosi" is now part of the systemd
14092 repository. mkosi is a tool to easily build legacy-free OS images,
14093 and is available on github: https://github.com/systemd/mkosi. If
14094 "mkosi" is invoked in the build tree a new raw OS image is generated
14095 incorporating the systemd sources currently being worked on and a
14096 clean, fresh distribution installation. The generated OS image may be
14097 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
14098 UEFI PC. This functionality is particularly useful to easily test
14099 local changes made to systemd in a pristine, defined environment. See
14100 doc/HACKING for details.
14102 * configure learned the --with-support-url= option to specify the
14103 distribution's bugtracker.
14105 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
14106 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
14107 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
14108 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
14109 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
14110 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
14111 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
14112 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
14113 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
14114 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
14115 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
14116 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
14117 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
14118 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
14119 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
14120 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
14121 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
14122 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
14123 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
14125 — Berlin, 2016-07-25
14129 * DNSSEC is now turned on by default in systemd-resolved (in
14130 "allow-downgrade" mode), but may be turned off during compile time by
14131 passing "--with-default-dnssec=no" to "configure" (and of course,
14132 during runtime with DNSSEC= in resolved.conf). We recommend
14133 downstreams to leave this on at least during development cycles and
14134 report any issues with the DNSSEC logic upstream. We are very
14135 interested in collecting feedback about the DNSSEC validator and its
14136 limitations in the wild. Note however, that DNSSEC support is
14137 probably nothing downstreams should turn on in stable distros just
14138 yet, as it might create incompatibilities with a few DNS servers and
14139 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
14140 automatically whenever we detect such incompatible setups, but there
14141 might be systems we do not cover yet. Hence: please help us testing
14142 the DNSSEC code, leave this on where you can, report back, but then
14143 again don't consider turning this on in your stable, LTS or
14144 production release just yet. (Note that you have to enable
14145 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
14146 and its DNSSEC mode for hostname resolution from local
14149 * systemd-resolve conveniently resolves DANE records with the --tlsa
14150 option and OPENPGPKEY records with the --openpgp option. It also
14151 supports dumping raw DNS record data via the new --raw= switch.
14153 * systemd-logind will now by default terminate user processes that are
14154 part of the user session scope unit (session-XX.scope) when the user
14155 logs out. This behavior is controlled by the KillUserProcesses=
14156 setting in logind.conf, and the previous default of "no" is now
14157 changed to "yes". This means that user sessions will be properly
14158 cleaned up after, but additional steps are necessary to allow
14159 intentionally long-running processes to survive logout.
14161 While the user is logged in at least once, user@.service is running,
14162 and any service that should survive the end of any individual login
14163 session can be started at a user service or scope using systemd-run.
14164 systemd-run(1) man page has been extended with an example which shows
14165 how to run screen in a scope unit underneath user@.service. The same
14166 command works for tmux.
14168 After the user logs out of all sessions, user@.service will be
14169 terminated too, by default, unless the user has "lingering" enabled.
14170 To effectively allow users to run long-term tasks even if they are
14171 logged out, lingering must be enabled for them. See loginctl(1) for
14172 details. The default polkit policy was modified to allow users to
14173 set lingering for themselves without authentication.
14175 Previous defaults can be restored at compile time by the
14176 --without-kill-user-processes option to "configure".
14178 * systemd-logind gained new configuration settings SessionsMax= and
14179 InhibitorsMax=, both with a default of 8192. It will not register new
14180 user sessions or inhibitors above this limit.
14182 * systemd-logind will now reload configuration on SIGHUP.
14184 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
14185 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
14186 enable. Also, support for the "io" cgroup controller in the unified
14187 hierarchy has been added, so that the "memory", "pids" and "io" are
14188 now the controllers that are supported on the unified hierarchy.
14190 WARNING: it is not possible to use previous systemd versions with
14191 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
14192 is necessary to also update systemd in the initramfs if using the
14193 unified hierarchy. An updated SELinux policy is also required.
14195 * LLDP support has been extended, and both passive (receive-only) and
14196 active (sender) modes are supported. Passive mode ("routers-only") is
14197 enabled by default in systemd-networkd. Active LLDP mode is enabled
14198 by default for containers on the internal network. The "networkctl
14199 lldp" command may be used to list information gathered. "networkctl
14200 status" will also show basic LLDP information on connected peers now.
14202 * The IAID and DUID unique identifier sent in DHCP requests may now be
14203 configured for the system and each .network file managed by
14204 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
14206 * systemd-networkd gained support for configuring proxy ARP support for
14207 each interface, via the ProxyArp= setting in .network files. It also
14208 gained support for configuring the multicast querier feature of
14209 bridge devices, via the new MulticastQuerier= setting in .netdev
14210 files. Similarly, snooping on the IGMP traffic can be controlled
14211 via the new setting MulticastSnooping=.
14213 A new setting PreferredLifetime= has been added for addresses
14214 configured in .network file to configure the lifetime intended for an
14217 The systemd-networkd DHCP server gained the option EmitRouter=, which
14218 defaults to yes, to configure whether the DHCP Option 3 (Router)
14221 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
14222 systemd-socket-activate and installed into /usr/bin. It is now fully
14225 * systemd-journald now uses separate threads to flush changes to disk
14226 when closing journal files, thus reducing impact of slow disk I/O on
14227 logging performance.
14229 * The sd-journal API gained two new calls
14230 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
14231 can be used to open journal files using file descriptors instead of
14232 file or directory paths. sd_journal_open_container() has been
14233 deprecated, sd_journal_open_directory_fd() should be used instead
14234 with the flag SD_JOURNAL_OS_ROOT.
14236 * journalctl learned a new output mode "-o short-unix" that outputs log
14237 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
14238 UTC). It also gained support for a new --no-hostname setting to
14239 suppress the hostname column in the family of "short" output modes.
14241 * systemd-ask-password now optionally skips printing of the password to
14242 stdout with --no-output which can be useful in scripts.
14244 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
14245 (devices tagged with ID_MAKER_TOOL) are now tagged with
14246 "uaccess" and are available to logged in users.
14248 * The DeviceAllow= unit setting now supports specifiers (with "%").
14250 * "systemctl show" gained a new --value switch, which allows print a
14251 only the contents of a specific unit property, without also printing
14252 the property's name. Similar support was added to "show*" verbs
14253 of loginctl and machinectl that output "key=value" lists.
14255 * A new unit type "generated" was added for files dynamically generated
14256 by generator tools. Similarly, a new unit type "transient" is used
14257 for unit files created using the runtime API. "systemctl enable" will
14258 refuse to operate on such files.
14260 * A new command "systemctl revert" has been added that may be used to
14261 revert to the vendor version of a unit file, in case local changes
14262 have been made by adding drop-ins or overriding the unit file.
14264 * "machinectl clean" gained a new verb to automatically remove all or
14265 just hidden container images.
14267 * systemd-tmpfiles gained support for a new line type "e" for emptying
14268 directories, if they exist, without creating them if they don't.
14270 * systemd-nspawn gained support for automatically patching the UID/GIDs
14271 of the owners and the ACLs of all files and directories in a
14272 container tree to match the UID/GID user namespacing range selected
14273 for the container invocation. This mode is enabled via the new
14274 --private-users-chown switch. It also gained support for
14275 automatically choosing a free, previously unused UID/GID range when
14276 starting a container, via the new --private-users=pick setting (which
14277 implies --private-users-chown). Together, these options for the first
14278 time make user namespacing for nspawn containers fully automatic and
14279 thus deployable. The systemd-nspawn@.service template unit file has
14280 been changed to use this functionality by default.
14282 * systemd-nspawn gained a new --network-zone= switch, that allows
14283 creating ad-hoc virtual Ethernet links between multiple containers,
14284 that only exist as long as at least one container referencing them is
14285 running. This allows easy connecting of multiple containers with a
14286 common link that implements an Ethernet broadcast domain. Each of
14287 these network "zones" may be named relatively freely by the user, and
14288 may be referenced by any number of containers, but each container may
14289 only reference one of these "zones". On the lower level, this is
14290 implemented by an automatically managed bridge network interface for
14291 each zone, that is created when the first container referencing its
14292 zone is created and removed when the last one referencing its zone
14295 * The default start timeout may now be configured on the kernel command
14296 line via systemd.default_timeout_start_sec=. It was already
14297 configurable via the DefaultTimeoutStartSec= option in
14298 /etc/systemd/system.conf.
14300 * Socket units gained a new TriggerLimitIntervalSec= and
14301 TriggerLimitBurst= setting to configure a limit on the activation
14302 rate of the socket unit.
14304 * The LimitNICE= setting now optionally takes normal UNIX nice values
14305 in addition to the raw integer limit value. If the specified
14306 parameter is prefixed with "+" or "-" and is in the range -20…19 the
14307 value is understood as UNIX nice value. If not prefixed like this it
14308 is understood as raw RLIMIT_NICE limit.
14310 * Note that the effect of the PrivateDevices= unit file setting changed
14311 slightly with this release: the per-device /dev file system will be
14312 mounted read-only from this version on, and will have "noexec"
14313 set. This (minor) change of behavior might cause some (exceptional)
14314 legacy software to break, when PrivateDevices=yes is set for its
14315 service. Please leave PrivateDevices= off if you run into problems
14318 * systemd-bootchart has been split out to a separate repository:
14319 https://github.com/systemd/systemd-bootchart
14321 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
14322 merged into the kernel in its current form.
14324 * The compatibility libraries libsystemd-daemon.so,
14325 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
14326 which have been deprecated since systemd-209 have been removed along
14327 with the corresponding pkg-config files. All symbols provided by
14328 those libraries are provided by libsystemd.so.
14330 * The Capabilities= unit file setting has been removed (it is ignored
14331 for backwards compatibility). AmbientCapabilities= and
14332 CapabilityBoundingSet= should be used instead.
14334 * A new special target has been added, initrd-root-device.target,
14335 which creates a synchronization point for dependencies of the root
14336 device in early userspace. Initramfs builders must ensure that this
14337 target is now included in early userspace.
14339 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
14340 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
14341 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
14342 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
14343 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
14344 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
14345 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
14346 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
14347 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
14348 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
14349 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
14350 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
14351 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
14352 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
14353 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
14354 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
14355 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
14356 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
14357 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
14358 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
14359 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
14360 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
14361 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
14362 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
14363 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
14366 — Fairfax, 2016-05-21
14370 * The systemd-resolved DNS resolver service has gained a substantial
14371 set of new features, most prominently it may now act as a DNSSEC
14372 validating stub resolver. DNSSEC mode is currently turned off by
14373 default, but is expected to be turned on by default in one of the
14374 next releases. For now, we invite everybody to test the DNSSEC logic
14375 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
14376 service also gained a full set of D-Bus interfaces, including calls
14377 to configure DNS and DNSSEC settings per link (for use by external
14378 network management software). systemd-resolved and systemd-networkd
14379 now distinguish between "search" and "routing" domains. The former
14380 are used to qualify single-label names, the latter are used purely
14381 for routing lookups within certain domains to specific links.
14382 resolved now also synthesizes RRs for all entries from /etc/hosts.
14384 * The systemd-resolve tool (which is a client utility for
14385 systemd-resolved) has been improved considerably and is now fully
14386 supported and documented. Hence it has moved from /usr/lib/systemd to
14389 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
14392 * The coredump collection logic has been reworked: when a coredump is
14393 collected it is now written to disk, compressed and processed
14394 (including stacktrace extraction) from a new instantiated service
14395 systemd-coredump@.service, instead of directly from the
14396 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
14397 processing large coredumps can take up a substantial amount of
14398 resources and time, and this previously happened entirely outside of
14399 systemd's service supervision. With the new logic the core_pattern
14400 hook only does minimal metadata collection before passing off control
14401 to the new instantiated service, which is configured with a time
14402 limit, a nice level and other settings to minimize negative impact on
14403 the rest of the system. Also note that the new logic will honour the
14404 RLIMIT_CORE setting of the crashed process, which now allows users
14405 and processes to turn off coredumping for their processes by setting
14408 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
14409 and all forked processes by default. Previously, PID 1 would leave
14410 the setting at "0" for all processes, as set by the kernel. Note that
14411 the resource limit traditionally has no effect on the generated
14412 coredumps on the system if the /proc/sys/kernel/core_pattern hook
14413 logic is used. Since the limit is now honoured (see above) its
14414 default has been changed so that the coredumping logic is enabled by
14415 default for all processes, while allowing specific opt-out.
14417 * When the stacktrace is extracted from processes of system users, this
14418 is now done as "systemd-coredump" user, in order to sandbox this
14419 potentially security sensitive parsing operation. (Note that when
14420 processing coredumps of normal users this is done under the user ID
14421 of process that crashed, as before.) Packagers should take notice
14422 that it is now necessary to create the "systemd-coredump" system user
14423 and group at package installation time.
14425 * The systemd-activate socket activation testing tool gained support
14426 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
14427 and --seqpacket switches. It also has been extended to support both
14428 new-style and inetd-style file descriptor passing. Use the new
14429 --inetd switch to request inetd-style file descriptor passing.
14431 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
14432 variable, which takes a boolean value. If set to false, ANSI color
14433 output is disabled in the tools even when run on a terminal that
14436 * The VXLAN support in networkd now supports two new settings
14437 DestinationPort= and PortRange=.
14439 * A new systemd.machine_id= kernel command line switch has been added,
14440 that may be used to set the machine ID in /etc/machine-id if it is
14441 not initialized yet. This command line option has no effect if the
14442 file is already initialized.
14444 * systemd-nspawn gained a new --as-pid2 switch that invokes any
14445 specified command line as PID 2 rather than PID 1 in the
14446 container. In this mode PID 1 is a minimal stub init process that
14447 implements the special POSIX and Linux semantics of PID 1 regarding
14448 signal and child process management. Note that this stub init process
14449 is implemented in nspawn itself and requires no support from the
14450 container image. This new logic is useful to support running
14451 arbitrary commands in the container, as normal processes are
14452 generally not prepared to run as PID 1.
14454 * systemd-nspawn gained a new --chdir= switch for setting the current
14455 working directory for the process started in the container.
14457 * "journalctl /dev/sda" will now output all kernel log messages for
14458 specified device from the current boot, in addition to all devices
14459 that are parents of it. This should make log output about devices
14460 pretty useful, as long as kernel drivers attach enough metadata to
14461 the log messages. (The usual SATA drivers do.)
14463 * The sd-journal API gained two new calls
14464 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
14465 that report whether log data from /run or /var has been found.
14467 * journalctl gained a new switch "--fields" that prints all journal
14468 record field names currently in use in the journal. This is backed
14469 by two new sd-journal API calls sd_journal_enumerate_fields() and
14470 sd_journal_restart_fields().
14472 * Most configurable timeouts in systemd now expect an argument of
14473 "infinity" to turn them off, instead of "0" as before. The semantics
14474 from now on is that a timeout of "0" means "now", and "infinity"
14475 means "never". To maintain backwards compatibility, "0" continues to
14476 turn off previously existing timeout settings.
14478 * "systemctl reload-or-try-restart" has been renamed to "systemctl
14479 try-reload-or-restart" to clarify what it actually does: the "try"
14480 logic applies to both reloading and restarting, not just restarting.
14481 The old name continues to be accepted for compatibility.
14483 * On boot-up, when PID 1 detects that the system clock is behind the
14484 release date of the systemd version in use, the clock is now set
14485 to the latter. Previously, this was already done in timesyncd, in order
14486 to avoid running with clocks set to the various clock epochs such as
14487 1902, 1938 or 1970. With this change the logic is now done in PID 1
14488 in addition to timesyncd during early boot-up, so that it is enforced
14489 before the first process is spawned by systemd. Note that the logic
14490 in timesyncd remains, as it is more comprehensive and ensures
14491 clock monotonicity by maintaining a persistent timestamp file in
14492 /var. Since /var is generally not available in earliest boot or the
14493 initrd, this part of the logic remains in timesyncd, and is not done
14496 * Support for tweaking details in net_cls.class_id through the
14497 NetClass= configuration directive has been removed, as the kernel
14498 people have decided to deprecate that controller in cgroup v2.
14499 Userspace tools such as nftables are moving over to setting rules
14500 that are specific to the full cgroup path of a task, which obsoletes
14501 these controllers anyway. The NetClass= directive is kept around for
14502 legacy compatibility reasons. For a more in-depth description of the
14503 kernel change, please refer to the respective upstream commit:
14505 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
14507 * A new service setting RuntimeMaxSec= has been added that may be used
14508 to specify a maximum runtime for a service. If the timeout is hit, the
14509 service is terminated and put into a failure state.
14511 * A new service setting AmbientCapabilities= has been added. It allows
14512 configuration of additional Linux process capabilities that are
14513 passed to the activated processes. This is only available on very
14516 * The process resource limit settings in service units may now be used
14517 to configure hard and soft limits individually.
14519 * The various libsystemd APIs such as sd-bus or sd-event now publicly
14520 expose support for gcc's __attribute__((cleanup())) C extension.
14521 Specifically, for many object destructor functions alternative
14522 versions have been added that have names suffixed with "p" and take a
14523 pointer to a pointer to the object to destroy, instead of just a
14524 pointer to the object itself. This is useful because these destructor
14525 functions may be used directly as parameters to the cleanup
14526 construct. Internally, systemd has been a heavy user of this GCC
14527 extension for a long time, and with this change similar support is
14528 now available to consumers of the library outside of systemd. Note
14529 that by using this extension in your sources compatibility with old
14530 and strictly ANSI compatible C compilers is lost. However, all gcc or
14531 LLVM versions of recent years support this extension.
14533 * Timer units gained support for a new setting RandomizedDelaySec= that
14534 allows configuring some additional randomized delay to the configured
14535 time. This is useful to spread out timer events to avoid load peaks in
14536 clusters or larger setups.
14538 * Calendar time specifications now support sub-second accuracy.
14540 * Socket units now support listening on SCTP and UDP-lite protocol
14543 * The sd-event API now comes with a full set of man pages.
14545 * Older versions of systemd contained experimental support for
14546 compressing journal files and coredumps with the LZ4 compressor that
14547 was not compatible with the lz4 binary (due to API limitations of the
14548 lz4 library). This support has been removed; only support for files
14549 compatible with the lz4 binary remains. This LZ4 logic is now
14550 officially supported and no longer considered experimental.
14552 * The dkr image import logic has been removed again from importd. dkr's
14553 micro-services focus doesn't fit into the machine image focus of
14554 importd, and quickly got out of date with the upstream dkr API.
14556 * Creation of the /run/lock/lockdev/ directory was dropped from
14557 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
14558 been available for many years. If you still need this, you need to
14559 create your own tmpfiles.d config file with:
14561 d /run/lock/lockdev 0775 root lock -
14563 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
14564 and RebootArgument= have been moved from the [Service] section of
14565 unit files to [Unit], and they are now supported on all unit types,
14566 not just service units. Of course, systemd will continue to
14567 understand these settings also at the old location, in order to
14568 maintain compatibility.
14570 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
14571 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
14572 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
14573 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
14574 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
14575 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
14576 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
14577 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
14578 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
14579 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
14580 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
14581 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
14582 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
14583 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
14584 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
14585 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
14586 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
14587 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
14588 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
14590 — Berlin, 2016-02-11
14594 * A number of properties previously only settable in unit
14595 files are now also available as properties to set when
14596 creating transient units programmatically via the bus, as it
14597 is exposed with systemd-run's --property=
14598 setting. Specifically, these are: SyslogIdentifier=,
14599 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
14600 EnvironmentFile=, ReadWriteDirectories=,
14601 ReadOnlyDirectories=, InaccessibleDirectories=,
14602 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
14604 * When creating transient services via the bus API it is now
14605 possible to pass in a set of file descriptors to use as
14606 STDIN/STDOUT/STDERR for the invoked process.
14608 * Slice units may now be created transiently via the bus APIs,
14609 similar to the way service and scope units may already be
14610 created transiently.
14612 * Wherever systemd expects a calendar timestamp specification
14613 (like in journalctl's --since= and --until= switches) UTC
14614 timestamps are now supported. Timestamps suffixed with "UTC"
14615 are now considered to be in Universal Time Coordinated
14616 instead of the local timezone. Also, timestamps may now
14617 optionally be specified with sub-second accuracy. Both of
14618 these additions also apply to recurring calendar event
14619 specification, such as OnCalendar= in timer units.
14621 * journalctl gained a new "--sync" switch that asks the
14622 journal daemon to write all so far unwritten log messages to
14623 disk and sync the files, before returning.
14625 * systemd-tmpfiles learned two new line types "q" and "Q" that
14626 operate like "v", but also set up a basic btrfs quota
14627 hierarchy when used on a btrfs file system with quota
14630 * tmpfiles' "v", "q" and "Q" will now create a plain directory
14631 instead of a subvolume (even on a btrfs file system) if the
14632 root directory is a plain directory, and not a
14633 subvolume. This should simplify things with certain chroot()
14634 environments which are not aware of the concept of btrfs
14637 * systemd-detect-virt gained a new --chroot switch to detect
14638 whether execution takes place in a chroot() environment.
14640 * CPUAffinity= now takes CPU index ranges in addition to
14641 individual indexes.
14643 * The various memory-related resource limit settings (such as
14644 LimitAS=) now understand the usual K, M, G, … suffixes to
14645 the base of 1024 (IEC). Similar, the time-related resource
14646 limit settings understand the usual min, h, day, … suffixes
14649 * There's a new system.conf setting DefaultTasksMax= to
14650 control the default TasksMax= setting for services and
14651 scopes running on the system. (TasksMax= is the primary
14652 setting that exposes the "pids" cgroup controller on systemd
14653 and was introduced in the previous systemd release.) The
14654 setting now defaults to 512, which means services that are
14655 not explicitly configured otherwise will only be able to
14656 create 512 processes or threads at maximum, from this
14657 version on. Note that this means that thread- or
14658 process-heavy services might need to be reconfigured to set
14659 TasksMax= to a higher value. It is sufficient to set
14660 TasksMax= in these specific unit files to a higher value, or
14661 even "infinity". Similar, there's now a logind.conf setting
14662 UserTasksMax= that defaults to 4096 and limits the total
14663 number of processes or tasks each user may own
14664 concurrently. nspawn containers also have the TasksMax=
14665 value set by default now, to 8192. Note that all of this
14666 only has an effect if the "pids" cgroup controller is
14667 enabled in the kernel. The general benefit of these changes
14668 should be a more robust and safer system, that provides a
14669 certain amount of per-service fork() bomb protection.
14671 * systemd-nspawn gained the new --network-veth-extra= switch
14672 to define additional and arbitrarily-named virtual Ethernet
14673 links between the host and the container.
14675 * A new service execution setting PassEnvironment= has been
14676 added that allows importing select environment variables
14677 from PID1's environment block into the environment block of
14680 * Timer units gained support for a new RemainAfterElapse=
14681 setting which takes a boolean argument. It defaults to on,
14682 exposing behaviour unchanged to previous releases. If set to
14683 off, timer units are unloaded after they elapsed if they
14684 cannot elapse again. This is particularly useful for
14685 transient timer units, which shall not stay around longer
14686 than until they first elapse.
14688 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
14689 default now (the kernel default is 16). This is beneficial
14690 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
14691 allows substantially larger numbers of queued
14692 datagrams. This should increase the capability of systemd to
14693 parallelize boot-up, as logging and sd_notify() are unlikely
14694 to stall execution anymore. If you need to change the value
14695 from the new defaults, use the usual sysctl.d/ snippets.
14697 * The compression framing format used by the journal or
14698 coredump processing has changed to be in line with what the
14699 official LZ4 tools generate. LZ4 compression support in
14700 systemd was considered unsupported previously, as the format
14701 was not compatible with the normal tools. With this release
14702 this has changed now, and it is hence safe for downstream
14703 distributions to turn it on. While not compressing as well
14704 as the XZ, LZ4 is substantially faster, which makes
14705 it a good default choice for the compression logic in the
14706 journal and in coredump handling.
14708 * Any reference to /etc/mtab has been dropped from
14709 systemd. The file has been obsolete since a while, but
14710 systemd refused to work on systems where it was incorrectly
14711 set up (it should be a symlink or non-existent). Please make
14712 sure to update to util-linux 2.27.1 or newer in conjunction
14713 with this systemd release, which also drops any reference to
14714 /etc/mtab. If you maintain a distribution make sure that no
14715 software you package still references it, as this is a
14716 likely source of bugs. There's also a glibc bug pending,
14717 asking for removal of any reference to this obsolete file:
14719 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
14721 Note that only util-linux versions built with
14722 --enable-libmount-force-mountinfo are supported.
14724 * Support for the ".snapshot" unit type has been removed. This
14725 feature turned out to be little useful and little used, and
14726 has now been removed from the core and from systemctl.
14728 * The dependency types RequiresOverridable= and
14729 RequisiteOverridable= have been removed from systemd. They
14730 have been used only very sparingly to our knowledge and
14731 other options that provide a similar effect (such as
14732 systemctl --mode=ignore-dependencies) are much more useful
14733 and commonly used. Moreover, they were only half-way
14734 implemented as the option to control behaviour regarding
14735 these dependencies was never added to systemctl. By removing
14736 these dependency types the execution engine becomes a bit
14737 simpler. Unit files that use these dependencies should be
14738 changed to use the non-Overridable dependency types
14739 instead. In fact, when parsing unit files with these
14740 options, that's what systemd will automatically convert them
14741 too, but it will also warn, asking users to fix the unit
14742 files accordingly. Removal of these dependency types should
14743 only affect a negligible number of unit files in the wild.
14745 * Behaviour of networkd's IPForward= option changed
14746 (again). It will no longer maintain a per-interface setting,
14747 but propagate one way from interfaces where this is enabled
14748 to the global kernel setting. The global setting will be
14749 enabled when requested by a network that is set up, but
14750 never be disabled again. This change was made to make sure
14751 IPv4 and IPv6 behaviour regarding packet forwarding is
14752 similar (as the Linux IPv6 stack does not support
14753 per-interface control of this setting) and to minimize
14756 * In unit files the behaviour of %u, %U, %h, %s has
14757 changed. These specifiers will now unconditionally resolve
14758 to the various user database fields of the user that the
14759 systemd instance is running as, instead of the user
14760 configured in the specific unit via User=. Note that this
14761 effectively doesn't change much, as resolving of these
14762 specifiers was already turned off in the --system instance
14763 of systemd, as we cannot do NSS lookups from PID 1. In the
14764 --user instance of systemd these specifiers where correctly
14765 resolved, but hardly made any sense, since the user instance
14766 lacks privileges to do user switches anyway, and User= is
14767 hence useless. Moreover, even in the --user instance of
14768 systemd behaviour was awkward as it would only take settings
14769 from User= assignment placed before the specifier into
14770 account. In order to unify and simplify the logic around
14771 this the specifiers will now always resolve to the
14772 credentials of the user invoking the manager (which in case
14773 of PID 1 is the root user).
14775 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
14776 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
14777 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
14778 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
14779 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
14780 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
14781 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
14782 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
14783 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
14784 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
14785 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
14786 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
14787 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
14788 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
14791 — Berlin, 2015-11-18
14795 * systemd now depends on util-linux v2.27. More specifically,
14796 the newly added mount monitor feature in libmount now
14797 replaces systemd's former own implementation.
14799 * libmount mandates /etc/mtab not to be regular file, and
14800 systemd now enforces this condition at early boot.
14801 /etc/mtab has been deprecated and warned about for a very
14802 long time, so systems running systemd should already have
14803 stopped having this file around as anything else than a
14804 symlink to /proc/self/mounts.
14806 * Support for the "pids" cgroup controller has been added. It
14807 allows accounting the number of tasks in a cgroup and
14808 enforcing limits on it. This adds two new setting
14809 TasksAccounting= and TasksMax= to each unit, as well as a
14810 global option DefaultTasksAccounting=.
14812 * Support for the "net_cls" cgroup controller has been added.
14813 It allows assigning a net class ID to each task in the
14814 cgroup, which can then be used in firewall rules and traffic
14815 shaping configurations. Note that the kernel netfilter net
14816 class code does not currently work reliably for ingress
14817 packets on unestablished sockets.
14819 This adds a new config directive called NetClass= to CGroup
14820 enabled units. Allowed values are positive numbers for fixed
14821 assignments and "auto" for picking a free value
14824 * 'systemctl is-system-running' now returns 'offline' if the
14825 system is not booted with systemd. This command can now be
14826 used as a substitute for 'systemd-notify --booted'.
14828 * Watchdog timeouts have been increased to 3 minutes for all
14829 in-tree service files. Apparently, disk IO issues are more
14830 frequent than we hoped, and user reported >1 minute waiting
14833 * 'machine-id-commit' functionality has been merged into
14834 'machine-id-setup --commit'. The separate binary has been
14837 * The WorkingDirectory= directive in unit files may now be set
14838 to the special value '~'. In this case, the working
14839 directory is set to the home directory of the user
14840 configured in User=.
14842 * "machinectl shell" will now open the shell in the home
14843 directory of the selected user by default.
14845 * The CrashChVT= configuration file setting is renamed to
14846 CrashChangeVT=, following our usual logic of not
14847 abbreviating unnecessarily. The old directive is still
14848 supported for compat reasons. Also, this directive now takes
14849 an integer value between 1 and 63, or a boolean value. The
14850 formerly supported '-1' value for disabling stays around for
14853 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
14854 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
14855 RootDirectory= properties can now be set for transient
14858 * The systemd-analyze tool gained a new "set-log-target" verb
14859 to change the logging target the system manager logs to
14860 dynamically during runtime. This is similar to how
14861 "systemd-analyze set-log-level" already changes the log
14864 * In nspawn /sys is now mounted as tmpfs, with only a selected
14865 set of subdirectories mounted in from the real sysfs. This
14866 enhances security slightly, and is useful for ensuring user
14867 namespaces work correctly.
14869 * Support for USB FunctionFS activation has been added. This
14870 allows implementation of USB gadget services that are
14871 activated as soon as they are requested, so that they don't
14872 have to run continuously, similar to classic socket
14875 * The "systemctl exit" command now optionally takes an
14876 additional parameter that sets the exit code to return from
14877 the systemd manager when exiting. This is only relevant when
14878 running the systemd user instance, or when running the
14879 system instance in a container.
14881 * sd-bus gained the new API calls sd_bus_path_encode_many()
14882 and sd_bus_path_decode_many() that allow easy encoding and
14883 decoding of multiple identifier strings inside a D-Bus
14884 object path. Another new call sd_bus_default_flush_close()
14885 has been added to flush and close per-thread default
14888 * systemd-cgtop gained support for a -M/--machine= switch to
14889 show the control groups within a certain container only.
14891 * "systemctl kill" gained support for an optional --fail
14892 switch. If specified the requested operation will fail of no
14893 processes have been killed, because the unit had no
14894 processes attached, or similar.
14896 * A new systemd.crash_reboot=1 kernel command line option has
14897 been added that triggers a reboot after crashing. This can
14898 also be set through CrashReboot= in systemd.conf.
14900 * The RuntimeDirectory= setting now understands unit
14901 specifiers like %i or %f.
14903 * A new (still internal) library API sd-ipv4acd has been added,
14904 that implements address conflict detection for IPv4. It's
14905 based on code from sd-ipv4ll, and will be useful for
14906 detecting DHCP address conflicts.
14908 * File descriptors passed during socket activation may now be
14909 named. A new API sd_listen_fds_with_names() is added to
14910 access the names. The default names may be overridden,
14911 either in the .socket file using the FileDescriptorName=
14912 parameter, or by passing FDNAME= when storing the file
14913 descriptors using sd_notify().
14915 * systemd-networkd gained support for:
14917 - Setting the IPv6 Router Advertisement settings via
14918 IPv6AcceptRouterAdvertisements= in .network files.
14920 - Configuring the HelloTimeSec=, MaxAgeSec= and
14921 ForwardDelaySec= bridge parameters in .netdev files.
14923 - Configuring PreferredSource= for static routes in
14926 * The "ask-password" framework used to query for LUKS harddisk
14927 passwords or SSL passwords during boot gained support for
14928 caching passwords in the kernel keyring, if it is
14929 available. This makes sure that the user only has to type in
14930 a passphrase once if there are multiple objects to unlock
14931 with the same one. Previously, such password caching was
14932 available only when Plymouth was used; this moves the
14933 caching logic into the systemd codebase itself. The
14934 "systemd-ask-password" utility gained a new --keyname=
14935 switch to control which kernel keyring key to use for
14936 caching a password in. This functionality is also useful for
14937 enabling display managers such as gdm to automatically
14938 unlock the user's GNOME keyring if its passphrase, the
14939 user's password and the harddisk password are the same, if
14940 gdm-autologin is used.
14942 * When downloading tar or raw images using "machinectl
14943 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
14944 file is now also downloaded, if it is available and stored
14945 next to the image file.
14947 * Units of type ".socket" gained a new boolean setting
14948 Writable= which is only useful in conjunction with
14949 ListenSpecial=. If true, enables opening the specified
14950 special file in O_RDWR mode rather than O_RDONLY mode.
14952 * systemd-rfkill has been reworked to become a singleton
14953 service that is activated through /dev/rfkill on each rfkill
14954 state change and saves the settings to disk. This way,
14955 systemd-rfkill is now compatible with devices that exist
14956 only intermittendly, and even restores state if the previous
14957 system shutdown was abrupt rather than clean.
14959 * The journal daemon gained support for vacuuming old journal
14960 files controlled by the number of files that shall remain,
14961 in addition to the already existing control by size and by
14962 date. This is useful as journal interleaving performance
14963 degrades with too many separate journal files, and allows
14964 putting an effective limit on them. The new setting defaults
14965 to 100, but this may be changed by setting SystemMaxFiles=
14966 and RuntimeMaxFiles= in journald.conf. Also, the
14967 "journalctl" tool gained the new --vacuum-files= switch to
14968 manually vacuum journal files to leave only the specified
14969 number of files in place.
14971 * udev will now create /dev/disk/by-path links for ATA devices
14972 on kernels where that is supported.
14974 * Galician, Serbian, Turkish and Korean translations were added.
14976 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
14977 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
14978 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
14979 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
14980 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
14981 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
14982 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
14983 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
14984 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
14985 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
14986 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
14987 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
14988 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
14989 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
14990 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
14991 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
14992 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
14993 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
14995 — Berlin, 2015-10-07
14999 * The DHCP implementation of systemd-networkd gained a set of
15002 - The DHCP server now supports emitting DNS and NTP
15003 information. It may be enabled and configured via
15004 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
15005 and NTP information is enabled, but no servers are
15006 configured, the corresponding uplink information (if there
15007 is any) is propagated.
15009 - Server and client now support transmission and reception
15010 of timezone information. It can be configured via the
15011 newly introduced network options UseTimezone=,
15012 EmitTimezone=, and Timezone=. Transmission of timezone
15013 information is enabled between host and containers by
15014 default now: the container will change its local timezone
15015 to what the host has set.
15017 - Lease timeouts can now be configured via
15018 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
15020 - The DHCP server improved on the stability of
15021 leases. Clients are more likely to get the same lease
15022 information back, even if the server loses state.
15024 - The DHCP server supports two new configuration options to
15025 control the lease address pool metrics, PoolOffset= and
15028 * The encapsulation limit of tunnels in systemd-networkd may
15029 now be configured via 'EncapsulationLimit='. It allows
15030 modifying the maximum additional levels of encapsulation
15031 that are permitted to be prepended to a packet.
15033 * systemd now supports the concept of user buses replacing
15034 session buses, if used with dbus-1.10 (and enabled via dbus
15035 --enable-user-session). It previously only supported this on
15036 kdbus-enabled systems, and this release expands this to
15037 'dbus-daemon' systems.
15039 * systemd-networkd now supports predictable interface names
15040 for virtio devices.
15042 * systemd now optionally supports the new Linux kernel
15043 "unified" control group hierarchy. If enabled via the kernel
15044 command-line option 'systemd.unified_cgroup_hierarchy=1',
15045 systemd will try to mount the unified cgroup hierarchy
15046 directly on /sys/fs/cgroup. If not enabled, or not
15047 available, systemd will fall back to the legacy cgroup
15048 hierarchy setup, as before. Host system and containers can
15049 mix and match legacy and unified hierarchies as they
15050 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
15051 environment variable to individually select the hierarchy to
15052 use for executed containers. By default, nspawn will use the
15053 unified hierarchy for the containers if the host uses the
15054 unified hierarchy, and the legacy hierarchy otherwise.
15055 Please note that at this point the unified hierarchy is an
15056 experimental kernel feature and is likely to change in one
15057 of the next kernel releases. Therefore, it should not be
15058 enabled by default in downstream distributions yet. The
15059 minimum required kernel version for the unified hierarchy to
15060 work is 4.2. Note that when the unified hierarchy is used
15061 for the first time delegated access to controllers is
15062 safe. Because of this systemd-nspawn containers will get
15063 access to controllers now, as will systemd user
15064 sessions. This means containers and user sessions may now
15065 manage their own resources, partitioning up what the system
15068 * A new special scope unit "init.scope" has been introduced
15069 that encapsulates PID 1 of the system. It may be used to
15070 determine resource usage and enforce resource limits on PID
15071 1 itself. PID 1 hence moved out of the root of the control
15074 * The cgtop tool gained support for filtering out kernel
15075 threads when counting tasks in a control group. Also, the
15076 count of processes is now recursively summed up by
15077 default. Two options -k and --recursive= have been added to
15078 revert to old behaviour. The tool has also been updated to
15079 work correctly in containers now.
15081 * systemd-nspawn's --bind= and --bind-ro= options have been
15082 extended to allow creation of non-recursive bind mounts.
15084 * libsystemd gained two new calls sd_pid_get_cgroup() and
15085 sd_peer_get_cgroup() which return the control group path of
15086 a process or peer of a connected AF_UNIX socket. This
15087 function call is particularly useful when implementing
15088 delegated subtrees support in the control group hierarchy.
15090 * The "sd-event" event loop API of libsystemd now supports
15091 correct dequeuing of real-time signals, without losing
15094 * When systemd requests a polkit decision when managing units it
15095 will now add additional fields to the request, including unit
15096 name and desired operation. This enables more powerful polkit
15097 policies, that make decisions depending on these parameters.
15099 * nspawn learnt support for .nspawn settings files, that may
15100 accompany the image files or directories of containers, and
15101 may contain additional settings for the container. This is
15102 an alternative to configuring container parameters via the
15103 nspawn command line.
15105 Contributions from: Cristian Rodríguez, Daniel Mack, David
15106 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
15107 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
15108 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
15109 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
15110 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
15111 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
15112 Andersen, Tom Gundersen, Torstein Husebø
15114 — Berlin, 2015-09-08
15118 * machinectl gained a new verb 'shell' which opens a fresh
15119 shell on the target container or the host. It is similar to
15120 the existing 'login' command of machinectl, but spawns the
15121 shell directly without prompting for username or
15122 password. The pseudo machine '.host' now refers to the local
15123 host and is used by default. Hence, 'machinectl shell' can
15124 be used as replacement for 'su -' which spawns a session as
15125 a fresh systemd unit in a way that is fully isolated from
15126 the originating session.
15128 * systemd-networkd learned to cope with private-zone DHCP
15129 options and allows other programs to query the values.
15131 * SELinux access control when enabling/disabling units is no
15132 longer enforced with this release. The previous implementation
15133 was incorrect, and a new corrected implementation is not yet
15134 available. As unit file operations are still protected via
15135 polkit and D-Bus policy this is not a security problem. Yet,
15136 distributions which care about optimal SELinux support should
15137 probably not stabilize on this release.
15139 * sd-bus gained support for matches of type "arg0has=", that
15140 test for membership of strings in string arrays sent in bus
15143 * systemd-resolved now dumps the contents of its DNS and LLMNR
15144 caches to the logs on reception of the SIGUSR1 signal. This
15145 is useful to debug DNS behaviour.
15147 * The coredumpctl tool gained a new --directory= option to
15148 operate on journal files in a specific directory.
15150 * "systemctl reboot" and related commands gained a new
15151 "--message=" option which may be used to set a free-text
15152 wall message when shutting down or rebooting the
15153 system. This message is also logged, which is useful for
15154 figuring out the reason for a reboot or shutdown a
15157 * The "systemd-resolve-host" tool's -i switch now takes
15158 network interface numbers as alternative to interface names.
15160 * A new unit file setting for services has been introduced:
15161 UtmpMode= allows configuration of how precisely systemd
15162 handles utmp and wtmp entries for the service if this is
15163 enabled. This allows writing services that appear similar to
15164 user sessions in the output of the "w", "who", "last" and
15167 * systemd-resolved will now locally synthesize DNS resource
15168 records for the "localhost" and "gateway" domains as well as
15169 the local hostname. This should ensure that clients querying
15170 RRs via resolved will get similar results as those going via
15171 NSS, if nss-myhostname is enabled.
15173 Contributions from: Alastair Hughes, Alex Crawford, Daniel
15174 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
15175 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
15176 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
15177 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
15178 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
15179 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
15180 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
15181 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
15182 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
15183 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
15184 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
15186 — Berlin, 2015-08-27
15190 * The systemd-efi-boot-generator functionality was merged into
15191 systemd-gpt-auto-generator.
15193 * systemd-networkd now supports Group Policy for vxlan
15194 devices. It can be enabled via the new boolean configuration
15195 option called 'GroupPolicyExtension='.
15197 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
15198 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
15199 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
15201 — Berlin, 2015-07-31
15205 * The python-systemd code has been removed from the systemd repository.
15206 A new repository has been created which accommodates the code from
15207 now on, and we kindly ask distributions to create a separate package
15208 for this: https://github.com/systemd/python-systemd
15210 * The systemd daemon will now reload its main configuration
15211 (/etc/systemd/system.conf) on daemon-reload.
15213 * sd-dhcp now exposes vendor specific extensions via
15214 sd_dhcp_lease_get_vendor_specific().
15216 * systemd-networkd gained a number of new configuration options.
15218 - A new boolean configuration option for TAP devices called
15219 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
15220 device, thus allowing to send and receive GSO packets.
15222 - A new tunnel configuration option called 'CopyDSCP='.
15223 If enabled, the DSCP field of ip6 tunnels is copied into the
15224 decapsulated packet.
15226 - A set of boolean bridge configuration options were added.
15227 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
15228 and 'UnicastFlood=' are now parsed by networkd and applied to the
15229 respective bridge link device via the respective IFLA_BRPORT_*
15232 - A new string configuration option to override the hostname sent
15233 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
15234 is true, networkd will use the configured hostname instead of the
15235 system hostname when sending DHCP requests.
15237 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
15238 networkd will configure the IPv6 flow-label of the tunnel device
15239 according to RFC2460.
15241 - The 'macvtap' virtual network devices are now supported, similar to
15242 the already supported 'macvlan' devices.
15244 * systemd-resolved now implements RFC5452 to improve resilience against
15245 cache poisoning. Additionally, source port randomization is enabled
15246 by default to further protect against DNS spoofing attacks.
15248 * nss-mymachines now supports translating UIDs and GIDs of running
15249 containers with user-namespaces enabled. If a container 'foo'
15250 translates a host uid 'UID' to the container uid 'TUID', then
15251 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
15252 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
15253 mapped as 'vg-foo-TGID'.
15255 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
15256 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
15257 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
15258 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
15259 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
15260 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
15261 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
15262 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
15263 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
15264 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
15266 — Berlin, 2015-07-29
15270 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
15271 There are no known issues with current sysfs, and udev does not need
15272 or should be used to work around such bugs.
15274 * udev does no longer enable USB HID power management. Several reports
15275 indicate, that some devices cannot handle that setting.
15277 * The udev accelerometer helper was removed. The functionality
15278 is now fully included in iio-sensor-proxy. But this means,
15279 older iio-sensor-proxy versions will no longer provide
15280 accelerometer/orientation data with this systemd version.
15281 Please upgrade iio-sensor-proxy to version 1.0.
15283 * networkd gained a new configuration option IPv6PrivacyExtensions=
15284 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
15285 for Stateless Address") on selected networks.
15287 * For the sake of fewer build-time dependencies and less code in the
15288 main repository, the python bindings are about to be removed in the
15289 next release. A new repository has been created which accommodates
15290 the code from now on, and we kindly ask distributions to create a
15291 separate package for this. The removal will take place in v223.
15293 https://github.com/systemd/python-systemd
15295 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
15296 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
15297 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
15298 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
15299 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
15300 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
15301 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
15302 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
15303 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
15304 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
15306 — Berlin, 2015-07-07
15310 * The sd-bus.h and sd-event.h APIs have now been declared
15311 stable and have been added to the official interface of
15312 libsystemd.so. sd-bus implements an alternative D-Bus client
15313 library, that is relatively easy to use, very efficient and
15314 supports both classic D-Bus as well as kdbus as transport
15315 backend. sd-event is a generic event loop abstraction that
15316 is built around Linux epoll, but adds features such as event
15317 prioritization or efficient timer handling. Both APIs are good
15318 choices for C programs looking for a bus and/or event loop
15319 implementation that is minimal and does not have to be
15320 portable to other kernels.
15322 * kdbus support is no longer compile-time optional. It is now
15323 always built-in. However, it can still be disabled at
15324 runtime using the kdbus=0 kernel command line setting, and
15325 that setting may be changed to default to off, by specifying
15326 --disable-kdbus at build-time. Note though that the kernel
15327 command line setting has no effect if the kdbus.ko kernel
15328 module is not installed, in which case kdbus is (obviously)
15329 also disabled. We encourage all downstream distributions to
15330 begin testing kdbus by adding it to the kernel images in the
15331 development distributions, and leaving kdbus support in
15334 * The minimal required util-linux version has been bumped to
15337 * Support for chkconfig (--enable-chkconfig) was removed in
15338 favor of calling an abstraction tool
15339 /lib/systemd/systemd-sysv-install. This needs to be
15340 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
15341 in README for details.
15343 * If there's a systemd unit and a SysV init script for the
15344 same service name, and the user executes "systemctl enable"
15345 for it (or a related call), then this will now enable both
15346 (or execute the related operation on both), not just the
15349 * The libudev API documentation has been converted from gtkdoc
15352 * gudev has been removed from the systemd tree, it is now an
15355 * The systemd-cgtop tool learnt a new --raw switch to generate
15356 "raw" (machine parsable) output.
15358 * networkd's IPForwarding= .network file setting learnt the
15359 new setting "kernel", which ensures that networkd does not
15360 change the IP forwarding sysctl from the default kernel
15363 * The systemd-logind bus API now exposes a new boolean
15364 property "Docked" that reports whether logind considers the
15365 system "docked", i.e. connected to a docking station or not.
15367 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
15368 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
15369 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
15370 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
15371 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
15372 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
15373 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
15374 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
15375 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
15376 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
15377 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
15378 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
15379 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
15380 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
15381 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
15382 Fink, Zbigniew Jędrzejewski-Szmek
15384 — Berlin, 2015-06-19
15388 * The gudev library has been extracted into a separate repository
15389 available at: https://git.gnome.org/browse/libgudev/
15390 It is now managed as part of the Gnome project. Distributions
15391 are recommended to pass --disable-gudev to systemd and use
15392 gudev from the Gnome project instead. gudev is still included
15393 in systemd, for now. It will be removed soon, though. Please
15394 also see the announcement-thread on systemd-devel:
15395 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
15397 * systemd now exposes a CPUUsageNSec= property for each
15398 service unit on the bus, that contains the overall consumed
15399 CPU time of a service (the sum of what each process of the
15400 service consumed). This value is only available if
15401 CPUAccounting= is turned on for a service, and is then shown
15402 in the "systemctl status" output.
15404 * Support for configuring alternative mappings of the old SysV
15405 runlevels to systemd targets has been removed. They are now
15406 hardcoded in a way that runlevels 2, 3, 4 all map to
15407 multi-user.target and 5 to graphical.target (which
15408 previously was already the default behaviour).
15410 * The auto-mounter logic gained support for mount point
15411 expiry, using a new TimeoutIdleSec= setting in .automount
15412 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
15414 * The EFI System Partition (ESP) as mounted to /boot by
15415 systemd-efi-boot-generator will now be unmounted
15416 automatically after 2 minutes of not being used. This should
15417 minimize the risk of ESP corruptions.
15419 * New /etc/fstab options x-systemd.requires= and
15420 x-systemd.requires-mounts-for= are now supported to express
15421 additional dependencies for mounts. This is useful for
15422 journaling file systems that support external journal
15423 devices or overlay file systems that require underlying file
15424 systems to be mounted.
15426 * systemd does not support direct live-upgrades (via systemctl
15427 daemon-reexec) from versions older than v44 anymore. As no
15428 distribution we are aware of shipped such old versions in a
15429 stable release this should not be problematic.
15431 * When systemd forks off a new per-connection service instance
15432 it will now set the $REMOTE_ADDR environment variable to the
15433 remote IP address, and $REMOTE_PORT environment variable to
15434 the remote IP port. This behaviour is similar to the
15435 corresponding environment variables defined by CGI.
15437 * systemd-networkd gained support for uplink failure
15438 detection. The BindCarrier= option allows binding interface
15439 configuration dynamically to the link sense of other
15440 interfaces. This is useful to achieve behaviour like in
15443 * systemd-networkd gained support for configuring the DHCP
15444 client identifier to use when requesting leases.
15446 * systemd-networkd now has a per-network UseNTP= option to
15447 configure whether NTP server information acquired via DHCP
15448 is passed on to services like systemd-timesyncd.
15450 * systemd-networkd gained support for vti6 tunnels.
15452 * Note that systemd-networkd manages the sysctl variable
15453 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
15454 it is configured for since v219. The variable controls IP
15455 forwarding, and is a per-interface alternative to the global
15456 /proc/sys/net/ipv[46]/ip_forward. This setting is
15457 configurable in the IPForward= option, which defaults to
15458 "no". This means if networkd is used for an interface it is
15459 no longer sufficient to set the global sysctl option to turn
15460 on IP forwarding! Instead, the .network file option
15461 IPForward= needs to be turned on! Note that the
15462 implementation of this behaviour was broken in v219 and has
15463 been fixed in v220.
15465 * Many bonding and vxlan options are now configurable in
15468 * systemd-nspawn gained a new --property= setting to set unit
15469 properties for the container scope. This is useful for
15470 setting resource parameters (e.g. "CPUShares=500") on
15471 containers started from the command line.
15473 * systemd-nspawn gained a new --private-users= switch to make
15474 use of user namespacing available on recent Linux kernels.
15476 * systemd-nspawn may now be called as part of a shell pipeline
15477 in which case the pipes used for stdin and stdout are passed
15478 directly to the process invoked in the container, without
15479 indirection via a pseudo tty.
15481 * systemd-nspawn gained a new switch to control the UNIX
15482 signal to use when killing the init process of the container
15483 when shutting down.
15485 * systemd-nspawn gained a new --overlay= switch for mounting
15486 overlay file systems into the container using the new kernel
15489 * When a container image is imported via systemd-importd and
15490 the host file system is not btrfs, a loopback block device
15491 file is created in /var/lib/machines.raw with a btrfs file
15492 system inside. It is then mounted to /var/lib/machines to
15493 enable btrfs features for container management. The loopback
15494 file and btrfs file system is grown as needed when container
15495 images are imported via systemd-importd.
15497 * systemd-machined/systemd-importd gained support for btrfs
15498 quota, to enforce container disk space limits on disk. This
15499 is exposed in "machinectl set-limit".
15501 * systemd-importd now can import containers from local .tar,
15502 .raw and .qcow2 images, and export them to .tar and .raw. It
15503 can also import dkr v2 images now from the network (on top
15506 * systemd-importd gained support for verifying downloaded
15507 images with gpg2 (previously only gpg1 was supported).
15509 * systemd-machined, systemd-logind, systemd: most bus calls are
15510 now accessible to unprivileged processes via polkit. Also,
15511 systemd-logind will now allow users to kill their own sessions
15512 without further privileges or authorization.
15514 * systemd-shutdownd has been removed. This service was
15515 previously responsible for implementing scheduled shutdowns
15516 as exposed in /usr/bin/shutdown's time parameter. This
15517 functionality has now been moved into systemd-logind and is
15518 accessible via a bus interface.
15520 * "systemctl reboot" gained a new switch --firmware-setup that
15521 can be used to reboot into the EFI firmware setup, if that
15522 is available. systemd-logind now exposes an API on the bus
15523 to trigger such reboots, in case graphical desktop UIs want
15524 to cover this functionality.
15526 * "systemctl enable", "systemctl disable" and "systemctl mask"
15527 now support a new "--now" switch. If specified the units
15528 that are enabled will also be started, and the ones
15529 disabled/masked also stopped.
15531 * The Gummiboot EFI boot loader tool has been merged into
15532 systemd, and renamed to "systemd-boot". The bootctl tool has been
15533 updated to support systemd-boot.
15535 * An EFI kernel stub has been added that may be used to create
15536 kernel EFI binaries that contain not only the actual kernel,
15537 but also an initrd, boot splash, command line and OS release
15538 information. This combined binary can then be signed as a
15539 single image, so that the firmware can verify it all in one
15540 step. systemd-boot has special support for EFI binaries created
15541 like this and can extract OS release information from them
15542 and show them in the boot menu. This functionality is useful
15543 to implement cryptographically verified boot schemes.
15545 * Optional support has been added to systemd-fsck to pass
15546 fsck's progress report to an AF_UNIX socket in the file
15549 * udev will no longer create device symlinks for all block devices by
15550 default. A deny list for excluding special block devices from this
15551 logic has been turned into an allow list that requires picking block
15552 devices explicitly that require device symlinks.
15554 * A new (currently still internal) API sd-device.h has been
15555 added to libsystemd. This modernized API is supposed to
15556 replace libudev eventually. In fact, already much of libudev
15557 is now just a wrapper around sd-device.h.
15559 * A new hwdb database for storing metadata about pointing
15560 stick devices has been added.
15562 * systemd-tmpfiles gained support for setting file attributes
15563 similar to the "chattr" tool with new 'h' and 'H' lines.
15565 * systemd-journald will no longer unconditionally set the
15566 btrfs NOCOW flag on new journal files. This is instead done
15567 with tmpfiles snippet using the new 'h' line type. This
15568 allows easy disabling of this logic, by masking the
15569 journal-nocow.conf tmpfiles file.
15571 * systemd-journald will now translate audit message types to
15572 human-readable identifiers when writing them to the
15573 journal. This should improve readability of audit messages.
15575 * The LUKS logic gained support for the offset= and skip=
15576 options in /etc/crypttab, as previously implemented by
15579 * /usr/lib/os-release gained a new optional field VARIANT= for
15580 distributions that support multiple variants (such as a
15581 desktop edition, a server edition, …)
15583 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
15584 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
15585 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
15586 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
15587 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
15588 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
15589 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
15590 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
15591 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
15592 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
15593 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
15594 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
15595 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
15596 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
15597 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
15598 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
15599 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
15600 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
15601 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
15602 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
15603 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
15604 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
15605 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
15606 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
15607 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
15608 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
15609 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
15611 — Berlin, 2015-05-22
15615 * Introduce a new API "sd-hwdb.h" for querying the hardware
15616 metadata database. With this minimal interface one can query
15617 and enumerate the udev hwdb, decoupled from the old libudev
15618 library. libudev's interface for this is now only a wrapper
15619 around sd-hwdb. A new tool systemd-hwdb has been added to
15620 interface with and update the database.
15622 * When any of systemd's tools copies files (for example due to
15623 tmpfiles' C lines) a btrfs reflink will attempted first,
15624 before bytewise copying is done.
15626 * systemd-nspawn gained a new --ephemeral switch. When
15627 specified a btrfs snapshot is taken of the container's root
15628 directory, and immediately removed when the container
15629 terminates again. Thus, a container can be started whose
15630 changes never alter the container's root directory, and are
15631 lost on container termination. This switch can also be used
15632 for starting a container off the root file system of the
15633 host without affecting the host OS. This switch is only
15634 available on btrfs file systems.
15636 * systemd-nspawn gained a new --template= switch. It takes the
15637 path to a container tree to use as template for the tree
15638 specified via --directory=, should that directory be
15639 missing. This allows instantiating containers dynamically,
15640 on first run. This switch is only available on btrfs file
15643 * When a .mount unit refers to a mount point on which multiple
15644 mounts are stacked, and the .mount unit is stopped all of
15645 the stacked mount points will now be unmounted until no
15646 mount point remains.
15648 * systemd now has an explicit notion of supported and
15649 unsupported unit types. Jobs enqueued for unsupported unit
15650 types will now fail with an "unsupported" error code. More
15651 specifically .swap, .automount and .device units are not
15652 supported in containers, .busname units are not supported on
15653 non-kdbus systems. .swap and .automount are also not
15654 supported if their respective kernel compile time options
15657 * machinectl gained support for two new "copy-from" and
15658 "copy-to" commands for copying files from a running
15659 container to the host or vice versa.
15661 * machinectl gained support for a new "bind" command to bind
15662 mount host directories into local containers. This is
15663 currently only supported for nspawn containers.
15665 * networkd gained support for configuring bridge forwarding
15666 database entries (fdb) from .network files.
15668 * A new tiny daemon "systemd-importd" has been added that can
15669 download container images in tar, raw, qcow2 or dkr formats,
15670 and make them available locally in /var/lib/machines, so
15671 that they can run as nspawn containers. The daemon can GPG
15672 verify the downloads (not supported for dkr, since it has no
15673 provisions for verifying downloads). It will transparently
15674 decompress bz2, xz, gzip compressed downloads if necessary,
15675 and restore sparse files on disk. The daemon uses privilege
15676 separation to ensure the actual download logic runs with
15677 fewer privileges than the daemon itself. machinectl has
15678 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
15679 make the functionality of importd available to the
15680 user. With this in place the Fedora and Ubuntu "Cloud"
15681 images can be downloaded and booted as containers unmodified
15682 (the Fedora images lack the appropriate GPG signature files
15683 currently, so they cannot be verified, but this will change
15684 soon, hopefully). Note that downloading images is currently
15685 only fully supported on btrfs.
15687 * machinectl is now able to list container images found in
15688 /var/lib/machines, along with some metadata about sizes of
15689 disk and similar. If the directory is located on btrfs and
15690 quota is enabled, this includes quota display. A new command
15691 "image-status" has been added that shows additional
15692 information about images.
15694 * machinectl is now able to clone container images
15695 efficiently, if the underlying file system (btrfs) supports
15696 it, with the new "machinectl clone" command. It also
15697 gained commands for renaming and removing images, as well as
15698 marking them read-only or read-write (supported also on
15699 legacy file systems).
15701 * networkd gained support for collecting LLDP network
15702 announcements, from hardware that supports this. This is
15703 shown in networkctl output.
15705 * systemd-run gained support for a new -t (--pty) switch for
15706 invoking a binary on a pty whose input and output is
15707 connected to the invoking terminal. This allows executing
15708 processes as system services while interactively
15709 communicating with them via the terminal. Most interestingly
15710 this is supported across container boundaries. Invoking
15711 "systemd-run -t /bin/bash" is an alternative to running a
15712 full login session, the difference being that the former
15713 will not register a session, nor go through the PAM session
15716 * tmpfiles gained support for a new "v" line type for creating
15717 btrfs subvolumes. If the underlying file system is a legacy
15718 file system, this automatically degrades to creating a
15719 normal directory. Among others /var/lib/machines is now
15720 created like this at boot, should it be missing.
15722 * The directory /var/lib/containers/ has been deprecated and
15723 been replaced by /var/lib/machines. The term "machines" has
15724 been used in the systemd context as generic term for both
15725 VMs and containers, and hence appears more appropriate for
15726 this, as the directory can also contain raw images bootable
15729 * systemd-nspawn when invoked with -M but without --directory=
15730 or --image= is now capable of searching for the container
15731 root directory, subvolume or disk image automatically, in
15732 /var/lib/machines. systemd-nspawn@.service has been updated
15733 to make use of this, thus allowing it to be used for raw
15736 * A new machines.target unit has been introduced that is
15737 supposed to group all containers/VMs invoked as services on
15738 the system. systemd-nspawn@.service has been updated to
15739 integrate with that.
15741 * machinectl gained a new "start" command, for invoking a
15742 container as a service. "machinectl start foo" is mostly
15743 equivalent to "systemctl start systemd-nspawn@foo.service",
15744 but handles escaping in a nicer way.
15746 * systemd-nspawn will now mount most of the cgroupfs tree
15747 read-only into each container, with the exception of the
15748 container's own subtree in the name=systemd hierarchy.
15750 * journald now sets the special FS_NOCOW file flag for its
15751 journal files. This should improve performance on btrfs, by
15752 avoiding heavy fragmentation when journald's write-pattern
15753 is used on COW file systems. It degrades btrfs' data
15754 integrity guarantees for the files to the same levels as for
15755 ext3/ext4 however. This should be OK though as journald does
15756 its own data integrity checks and all its objects are
15757 checksummed on disk. Also, journald should handle btrfs disk
15758 full events a lot more gracefully now, by processing SIGBUS
15759 errors, and not relying on fallocate() anymore.
15761 * When journald detects that journal files it is writing to
15762 have been deleted it will immediately start new journal
15765 * systemd now provides a way to store file descriptors
15766 per-service in PID 1. This is useful for daemons to ensure
15767 that fds they require are not lost during a daemon
15768 restart. The fds are passed to the daemon on the next
15769 invocation in the same way socket activation fds are
15770 passed. This is now used by journald to ensure that the
15771 various sockets connected to all the system's stdout/stderr
15772 are not lost when journald is restarted. File descriptors
15773 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
15774 an extension to sd_notify(). Note that a limit is enforced
15775 on the number of fds a service can store in PID 1, and it
15776 defaults to 0, so that no fds may be stored, unless this is
15777 explicitly turned on.
15779 * The default TERM variable to use for units connected to a
15780 terminal, when no other value is explicitly is set is now
15781 vt220 rather than vt102. This should be fairly safe still,
15782 but allows PgUp/PgDn work.
15784 * The /etc/crypttab option header= as known from Debian is now
15787 * "loginctl user-status" and "loginctl session-status" will
15788 now show the last 10 lines of log messages of the
15789 user/session following the status output. Similar,
15790 "machinectl status" will show the last 10 log lines
15791 associated with a virtual machine or container
15792 service. (Note that this is usually not the log messages
15793 done in the VM/container itself, but simply what the
15794 container manager logs. For nspawn this includes all console
15797 * "loginctl session-status" without further argument will now
15798 show the status of the session of the caller. Similar,
15799 "lock-session", "unlock-session", "activate",
15800 "enable-linger", "disable-linger" may now be called without
15801 session/user parameter in which case they apply to the
15802 caller's session/user.
15804 * An X11 session scriptlet is now shipped that uploads
15805 $DISPLAY and $XAUTHORITY into the environment of the systemd
15806 --user daemon if a session begins. This should improve
15807 compatibility with X11 enabled applications run as systemd
15810 * Generators are now subject to masking via /etc and /run, the
15811 same way as unit files.
15813 * networkd .network files gained support for configuring
15814 per-link IPv4/IPv6 packet forwarding as well as IPv4
15815 masquerading. This is by default turned on for veth links to
15816 containers, as registered by systemd-nspawn. This means that
15817 nspawn containers run with --network-veth will now get
15818 automatic routed access to the host's networks without any
15819 further configuration or setup, as long as networkd runs on
15822 * systemd-nspawn gained the --port= (-p) switch to expose TCP
15823 or UDP posts of a container on the host. With this in place
15824 it is possible to run containers with private veth links
15825 (--network-veth), and have their functionality exposed on
15826 the host as if their services were running directly on the
15829 * systemd-nspawn's --network-veth switch now gained a short
15830 version "-n", since with the changes above it is now truly
15831 useful out-of-the-box. The systemd-nspawn@.service has been
15832 updated to make use of it too by default.
15834 * systemd-nspawn will now maintain a per-image R/W lock, to
15835 ensure that the same image is not started more than once
15836 writable. (It's OK to run an image multiple times
15837 simultaneously in read-only mode.)
15839 * systemd-nspawn's --image= option is now capable of
15840 dissecting and booting MBR and GPT disk images that contain
15841 only a single active Linux partition. Previously it
15842 supported only GPT disk images with proper GPT type
15843 IDs. This allows running cloud images from major
15844 distributions directly with systemd-nspawn, without
15847 * In addition to collecting mouse dpi data in the udev
15848 hardware database, there's now support for collecting angle
15849 information for mouse scroll wheels. The database is
15850 supposed to guarantee similar scrolling behavior on mice
15851 that it knows about. There's also support for collecting
15852 information about Touchpad types.
15854 * udev's input_id built-in will now also collect touch screen
15855 dimension data and attach it to probed devices.
15857 * /etc/os-release gained support for a Distribution Privacy
15860 * networkd gained support for creating "ipvlan", "gretap",
15861 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
15863 * systemd-tmpfiles gained support for "a" lines for setting
15866 * systemd-nspawn will now mount /tmp in the container to
15867 tmpfs, automatically.
15869 * systemd now exposes the memory.usage_in_bytes cgroup
15870 attribute and shows it for each service in the "systemctl
15871 status" output, if available.
15873 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
15874 immediate reboot is triggered. This useful if shutdown is
15875 hung and is unable to complete, to expedite the
15876 operation. Note that this kind of reboot will still unmount
15877 all file systems, and hence should not result in fsck being
15878 run on next reboot.
15880 * A .device unit for an optical block device will now be
15881 considered active only when a medium is in the drive. Also,
15882 mount units are now bound to their backing devices thus
15883 triggering automatic unmounting when devices become
15884 unavailable. With this in place systemd will now
15885 automatically unmount left-over mounts when a CD-ROM is
15886 ejected or a USB stick is yanked from the system.
15888 * networkd-wait-online now has support for waiting for
15889 specific interfaces only (with globbing), and for giving up
15890 after a configurable timeout.
15892 * networkd now exits when idle. It will be automatically
15893 restarted as soon as interfaces show up, are removed or
15894 change state. networkd will stay around as long as there is
15895 at least one DHCP state machine or similar around, that keep
15898 * networkd may now configure IPv6 link-local addressing in
15899 addition to IPv4 link-local addressing.
15901 * The IPv6 "token" for use in SLAAC may now be configured for
15902 each .network interface in networkd.
15904 * Routes configured with networkd may now be assigned a scope
15907 * networkd's [Match] sections now support globbing and lists
15908 of multiple space-separated matches per item.
15910 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
15911 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
15912 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
15913 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
15914 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
15915 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
15916 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
15917 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
15918 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
15919 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
15920 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
15921 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
15922 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
15923 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
15924 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
15925 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
15926 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
15927 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
15928 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
15929 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
15930 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
15931 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
15932 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
15933 Hoffmann, Zbigniew Jędrzejewski-Szmek
15935 — Berlin, 2015-02-16
15939 * When querying unit file enablement status (for example via
15940 "systemctl is-enabled"), a new state "indirect" is now known
15941 which indicates that a unit might not be enabled itself, but
15942 another unit listed in its Also= setting might be.
15944 * Similar to the various existing ConditionXYZ= settings for
15945 units, there are now matching AssertXYZ= settings. While
15946 failing conditions cause a unit to be skipped, but its job
15947 to succeed, failing assertions declared like this will cause
15948 a unit start operation and its job to fail.
15950 * hostnamed now knows a new chassis type "embedded".
15952 * systemctl gained a new "edit" command. When used on a unit
15953 file, this allows extending unit files with .d/ drop-in
15954 configuration snippets or editing the full file (after
15955 copying it from /usr/lib to /etc). This will invoke the
15956 user's editor (as configured with $EDITOR), and reload the
15957 modified configuration after editing.
15959 * "systemctl status" now shows the suggested enablement state
15960 for a unit, as declared in the (usually vendor-supplied)
15961 system preset files.
15963 * nss-myhostname will now resolve the single-label hostname
15964 "gateway" to the locally configured default IP routing
15965 gateways, ordered by their metrics. This assigns a stable
15966 name to the used gateways, regardless which ones are
15967 currently configured. Note that the name will only be
15968 resolved after all other name sources (if nss-myhostname is
15969 configured properly) and should hence not negatively impact
15970 systems that use the single-label hostname "gateway" in
15973 * systemd-inhibit now allows filtering by mode when listing
15976 * Scope and service units gained a new "Delegate" boolean
15977 property, which, when set, allows processes running inside the
15978 unit to further partition resources. This is primarily
15979 useful for systemd user instances as well as container
15982 * journald will now pick up audit messages directly from
15983 the kernel, and log them like any other log message. The
15984 audit fields are split up and fully indexed. This means that
15985 journalctl in many ways is now a (nicer!) alternative to
15986 ausearch, the traditional audit client. Note that this
15987 implements only a minimal audit client. If you want the
15988 special audit modes like reboot-on-log-overflow, please use
15989 the traditional auditd instead, which can be used in
15990 parallel to journald.
15992 * The ConditionSecurity= unit file option now understands the
15993 special string "audit" to check whether auditing is
15996 * journalctl gained two new commands --vacuum-size= and
15997 --vacuum-time= to delete old journal files until the
15998 remaining ones take up no more than the specified size on disk,
15999 or are not older than the specified time.
16001 * A new, native PPPoE library has been added to sd-network,
16002 systemd's library of lightweight networking protocols. This
16003 library will be used in a future version of networkd to
16004 enable PPPoE communication without an external pppd daemon.
16006 * The busctl tool now understands a new "capture" verb that
16007 works similar to "monitor", but writes a packet capture
16008 trace to STDOUT that can be redirected to a file which is
16009 compatible with libcap's capture file format. This can then
16010 be loaded in Wireshark and similar tools to inspect bus
16013 * The busctl tool now understands a new "tree" verb that shows
16014 the object trees of a specific service on the bus, or of all
16017 * The busctl tool now understands a new "introspect" verb that
16018 shows all interfaces and members of objects on the bus,
16019 including their signature and values. This is particularly
16020 useful to get more information about bus objects shown by
16021 the new "busctl tree" command.
16023 * The busctl tool now understands new verbs "call",
16024 "set-property" and "get-property" for invoking bus method
16025 calls, setting and getting bus object properties in a
16028 * busctl gained a new --augment-creds= argument that controls
16029 whether the tool shall augment credential information it
16030 gets from the bus with data from /proc, in a possibly
16033 * nspawn's --link-journal= switch gained two new values
16034 "try-guest" and "try-host" that work like "guest" and
16035 "host", but do not fail if the host has no persistent
16036 journaling enabled. -j is now equivalent to
16037 --link-journal=try-guest.
16039 * macvlan network devices created by nspawn will now have
16040 stable MAC addresses.
16042 * A new SmackProcessLabel= unit setting has been added, which
16043 controls the SMACK security label processes forked off by
16044 the respective unit shall use.
16046 * If compiled with --enable-xkbcommon, systemd-localed will
16047 verify x11 keymap settings by compiling the given keymap. It
16048 will spew out warnings if the compilation fails. This
16049 requires libxkbcommon to be installed.
16051 * When a coredump is collected, a larger number of metadata
16052 fields is now collected and included in the journal records
16053 created for it. More specifically, control group membership,
16054 environment variables, memory maps, working directory,
16055 chroot directory, /proc/$PID/status, and a list of open file
16056 descriptors is now stored in the log entry.
16058 * The udev hwdb now contains DPI information for mice. For
16061 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
16063 * All systemd programs that read standalone configuration
16064 files in /etc now also support a corresponding series of
16065 .conf.d configuration directories in /etc/, /run/,
16066 /usr/local/lib/, /usr/lib/, and (if configured with
16067 --enable-split-usr) /lib/. In particular, the following
16068 configuration files now have corresponding configuration
16069 directories: system.conf user.conf, logind.conf,
16070 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
16071 resolved.conf, timesyncd.conf, journal-remote.conf, and
16072 journal-upload.conf. Note that distributions should use the
16073 configuration directories in /usr/lib/; the directories in
16074 /etc/ are reserved for the system administrator.
16076 * systemd-rfkill will no longer take the rfkill device name
16077 into account when storing rfkill state on disk, as the name
16078 might be dynamically assigned and not stable. Instead, the
16079 ID_PATH udev variable combined with the rfkill type (wlan,
16080 bluetooth, …) is used.
16082 * A new service systemd-machine-id-commit.service has been
16083 added. When used on systems where /etc is read-only during
16084 boot, and /etc/machine-id is not initialized (but an empty
16085 file), this service will copy the temporary machine ID
16086 created as replacement into /etc after the system is fully
16087 booted up. This is useful for systems that are freshly
16088 installed with a non-initialized machine ID, but should get
16089 a fixed machine ID for subsequent boots.
16091 * networkd's .netdev files now provide a large set of
16092 configuration parameters for VXLAN devices. Similarly, the
16093 bridge port cost parameter is now configurable in .network
16094 files. There's also new support for configuring IP source
16095 routing. networkd .link files gained support for a new
16096 OriginalName= match that is useful to match against the
16097 original interface name the kernel assigned. .network files
16098 may include MTU= and MACAddress= fields for altering the MTU
16099 and MAC address while being connected to a specific network
16102 * The LUKS logic gained supported for configuring
16103 UUID-specific key files. There's also new support for naming
16104 LUKS device from the kernel command line, using the new
16105 luks.name= argument.
16107 * Timer units may now be transiently created via the bus API
16108 (this was previously already available for scope and service
16109 units). In addition it is now possible to create multiple
16110 transient units at the same time with a single bus call. The
16111 "systemd-run" tool has been updated to make use of this for
16112 running commands on a specified time, in at(1)-style.
16114 * tmpfiles gained support for "t" lines, for assigning
16115 extended attributes to files. Among other uses this may be
16116 used to assign SMACK labels to files.
16118 Contributions from: Alin Rauta, Alison Chaiken, Andrej
16119 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
16120 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
16121 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
16122 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
16123 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
16124 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
16125 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
16126 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
16127 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
16128 Michael Biebl, Michael Chapman, Michael Marineau, Michal
16129 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
16130 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
16131 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
16132 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
16133 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
16134 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
16135 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
16137 — Berlin, 2014-12-10
16141 * journalctl gained the new options -t/--identifier= to match
16142 on the syslog identifier (aka "tag"), as well as --utc to
16143 show log timestamps in the UTC timezone. journalctl now also
16144 accepts -n/--lines=all to disable line capping in a pager.
16146 * journalctl gained a new switch, --flush, that synchronously
16147 flushes logs from /run/log/journal to /var/log/journal if
16148 persistent storage is enabled. systemd-journal-flush.service
16149 now waits until the operation is complete.
16151 * Services can notify the manager before they start a reload
16152 (by sending RELOADING=1) or shutdown (by sending
16153 STOPPING=1). This allows the manager to track and show the
16154 internal state of daemons and closes a race condition when
16155 the process is still running but has closed its D-Bus
16158 * Services with Type=oneshot do not have to have any ExecStart
16161 * User units are now loaded also from
16162 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
16163 /run/systemd/user directory that was already previously
16164 supported, but is under the control of the user.
16166 * Job timeouts (i.e. timeouts on the time a job that is
16167 queued stays in the run queue) can now optionally result in
16168 immediate reboot or power-off actions (JobTimeoutAction= and
16169 JobTimeoutRebootArgument=). This is useful on ".target"
16170 units, to limit the maximum time a target remains
16171 undispatched in the run queue, and to trigger an emergency
16172 operation in such a case. This is now used by default to
16173 turn off the system if boot-up (as defined by everything in
16174 basic.target) hangs and does not complete for at least
16175 15min. Also, if power-off or reboot hang for at least 30min
16176 an immediate power-off/reboot operation is triggered. This
16177 functionality is particularly useful to increase reliability
16178 on embedded devices, but also on laptops which might
16179 accidentally get powered on when carried in a backpack and
16180 whose boot stays stuck in a hard disk encryption passphrase
16183 * systemd-logind can be configured to also handle lid switch
16184 events even when the machine is docked or multiple displays
16185 are attached (HandleLidSwitchDocked= option).
16187 * A helper binary and a service have been added which can be
16188 used to resume from hibernation in the initramfs. A
16189 generator will parse the resume= option on the kernel
16190 command line to trigger resume.
16192 * A user console daemon systemd-consoled has been
16193 added. Currently, it is a preview, and will so far open a
16194 single terminal on each session of the user marked as
16195 Desktop=systemd-console.
16197 * Route metrics can be specified for DHCP routes added by
16200 * The SELinux context of socket-activated services can be set
16201 from the information provided by the networking stack
16202 (SELinuxContextFromNet= option).
16204 * Userspace firmware loading support has been removed and
16205 the minimum supported kernel version is thus bumped to 3.7.
16207 * Timeout for udev workers has been increased from 1 to 3
16208 minutes, but a warning will be printed after 1 minute to
16209 help diagnose kernel modules that take a long time to load.
16211 * Udev rules can now remove tags on devices with TAG-="foobar".
16213 * systemd's readahead implementation has been removed. In many
16214 circumstances it didn't give expected benefits even for
16215 rotational disk drives and was becoming less relevant in the
16216 age of SSDs. As none of the developers has been using
16217 rotating media anymore, and nobody stepped up to actively
16218 maintain this component of systemd it has now been removed.
16220 * Swap units can use Options= to specify discard options.
16221 Discard options specified for swaps in /etc/fstab are now
16224 * Docker containers are now detected as a separate type of
16227 * The Password Agent protocol gained support for queries where
16228 the user input is shown, useful e.g. for user names.
16229 systemd-ask-password gained a new --echo option to turn that
16232 * The default sysctl.d/ snippets will now set:
16234 net.core.default_qdisc = fq_codel
16236 This selects Fair Queuing Controlled Delay as the default
16237 queuing discipline for network interfaces. fq_codel helps
16238 fight the network bufferbloat problem. It is believed to be
16239 a good default with no tuning required for most workloads.
16240 Downstream distributions may override this choice. On 10Gbit
16241 servers that do not do forwarding, "fq" may perform better.
16242 Systems without a good clocksource should use "pfifo_fast".
16244 * If kdbus is enabled during build a new option BusPolicy= is
16245 available for service units, that allows locking all service
16246 processes into a stricter bus policy, in order to limit
16247 access to various bus services, or even hide most of them
16248 from the service's view entirely.
16250 * networkctl will now show the .network and .link file
16251 networkd has applied to a specific interface.
16253 * sd-login gained a new API call sd_session_get_desktop() to
16254 query which desktop environment has been selected for a
16257 * UNIX utmp support is now compile-time optional to support
16258 legacy-free systems.
16260 * systemctl gained two new commands "add-wants" and
16261 "add-requires" for pulling in units from specific targets
16264 * If the word "rescue" is specified on the kernel command line
16265 the system will now boot into rescue mode (aka
16266 rescue.target), which was previously available only by
16267 specifying "1" or "systemd.unit=rescue.target" on the kernel
16268 command line. This new kernel command line option nicely
16269 mirrors the already existing "emergency" kernel command line
16272 * New kernel command line options mount.usr=, mount.usrflags=,
16273 mount.usrfstype= have been added that match root=, rootflags=,
16274 rootfstype= but allow mounting a specific file system to
16277 * The $NOTIFY_SOCKET is now also passed to control processes of
16278 services, not only the main process.
16280 * This version reenables support for fsck's -l switch. This
16281 means at least version v2.25 of util-linux is required for
16282 operation, otherwise dead-locks on device nodes may
16283 occur. Again: you need to update util-linux to at least
16284 v2.25 when updating systemd to v217.
16286 * The "multi-seat-x" tool has been removed from systemd, as
16287 its functionality has been integrated into X servers 1.16,
16288 and the tool is hence redundant. It is recommended to update
16289 display managers invoking this tool to simply invoke X
16290 directly from now on, again.
16292 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
16293 message flag has been added for all of systemd's polkit
16294 authenticated method calls has been added. In particular this
16295 now allows optional interactive authorization via polkit for
16296 many of PID1's privileged operations such as unit file
16297 enabling and disabling.
16299 * "udevadm hwdb --update" learnt a new switch "--usr" for
16300 placing the rebuilt hardware database in /usr instead of
16301 /etc. When used only hardware database entries stored in
16302 /usr will be used, and any user database entries in /etc are
16303 ignored. This functionality is useful for vendors to ship a
16304 pre-built database on systems where local configuration is
16305 unnecessary or unlikely.
16307 * Calendar time specifications in .timer units now also
16308 understand the strings "semi-annually", "quarterly" and
16309 "minutely" as shortcuts (in addition to the preexisting
16310 "annually", "hourly", …).
16312 * systemd-tmpfiles will now correctly create files in /dev
16313 at boot which are marked for creation only at boot. It is
16314 recommended to always create static device nodes with 'c!'
16315 and 'b!', so that they are created only at boot and not
16316 overwritten at runtime.
16318 * When the watchdog logic is used for a service (WatchdogSec=)
16319 and the watchdog timeout is hit the service will now be
16320 terminated with SIGABRT (instead of just SIGTERM), in order
16321 to make sure a proper coredump and backtrace is
16322 generated. This ensures that hanging services will result in
16323 similar coredump/backtrace behaviour as services that hit a
16324 segmentation fault.
16326 Contributions from: Andreas Henriksson, Andrei Borzenkov,
16327 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
16328 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
16329 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
16330 Herrmann, David Sommerseth, David Strauss, Emil Renner
16331 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
16332 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
16333 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
16334 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
16335 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
16336 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
16337 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
16338 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
16339 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
16340 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
16341 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
16342 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
16343 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
16344 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
16345 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
16346 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
16349 — Berlin, 2014-10-28
16353 * timedated no longer reads NTP implementation unit names from
16354 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
16355 implementations should add a
16357 Conflicts=systemd-timesyncd.service
16359 to their unit files to take over and replace systemd's NTP
16360 default functionality.
16362 * systemd-sysusers gained a new line type "r" for configuring
16363 which UID/GID ranges to allocate system users/groups
16364 from. Lines of type "u" may now add an additional column
16365 that specifies the home directory for the system user to be
16366 created. Also, systemd-sysusers may now optionally read user
16367 information from STDIN instead of a file. This is useful for
16368 invoking it from RPM preinst scriptlets that need to create
16369 users before the first RPM file is installed since these
16370 files might need to be owned by them. A new
16371 %sysusers_create_inline RPM macro has been introduced to do
16372 just that. systemd-sysusers now updates the shadow files as
16373 well as the user/group databases, which should enhance
16374 compatibility with certain tools like grpck.
16376 * A number of bus APIs of PID 1 now optionally consult polkit to
16377 permit access for otherwise unprivileged clients under certain
16378 conditions. Note that this currently doesn't support
16379 interactive authentication yet, but this is expected to be
16380 added eventually, too.
16382 * /etc/machine-info now has new fields for configuring the
16383 deployment environment of the machine, as well as the
16384 location of the machine. hostnamectl has been updated with
16385 new command to update these fields.
16387 * systemd-timesyncd has been updated to automatically acquire
16388 NTP server information from systemd-networkd, which might
16389 have been discovered via DHCP.
16391 * systemd-resolved now includes a caching DNS stub resolver
16392 and a complete LLMNR name resolution implementation. A new
16393 NSS module "nss-resolve" has been added which can be used
16394 instead of glibc's own "nss-dns" to resolve hostnames via
16395 systemd-resolved. Hostnames, addresses and arbitrary RRs may
16396 be resolved via systemd-resolved D-Bus APIs. In contrast to
16397 the glibc internal resolver systemd-resolved is aware of
16398 multi-homed system, and keeps DNS server and caches separate
16399 and per-interface. Queries are sent simultaneously on all
16400 interfaces that have DNS servers configured, in order to
16401 properly handle VPNs and local LANs which might resolve
16402 separate sets of domain names. systemd-resolved may acquire
16403 DNS server information from systemd-networkd automatically,
16404 which in turn might have discovered them via DHCP. A tool
16405 "systemd-resolve-host" has been added that may be used to
16406 query the DNS logic in resolved. systemd-resolved implements
16407 IDNA and automatically uses IDNA or UTF-8 encoding depending
16408 on whether classic DNS or LLMNR is used as transport. In the
16409 next releases we intend to add a DNSSEC and mDNS/DNS-SD
16410 implementation to systemd-resolved.
16412 * A new NSS module nss-mymachines has been added, that
16413 automatically resolves the names of all local registered
16414 containers to their respective IP addresses.
16416 * A new client tool "networkctl" for systemd-networkd has been
16417 added. It currently is entirely passive and will query
16418 networking configuration from udev, rtnetlink and networkd,
16419 and present it to the user in a very friendly
16420 way. Eventually, we hope to extend it to become a full
16421 control utility for networkd.
16423 * .socket units gained a new DeferAcceptSec= setting that
16424 controls the kernels' TCP_DEFER_ACCEPT sockopt for
16425 TCP. Similarly, support for controlling TCP keep-alive
16426 settings has been added (KeepAliveTimeSec=,
16427 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
16428 turning off Nagle's algorithm on TCP has been added
16431 * logind learned a new session type "web", for use in projects
16432 like Cockpit which register web clients as PAM sessions.
16434 * timer units with at least one OnCalendar= setting will now
16435 be started only after time-sync.target has been
16436 reached. This way they will not elapse before the system
16437 clock has been corrected by a local NTP client or
16438 similar. This is particular useful on RTC-less embedded
16439 machines, that come up with an invalid system clock.
16441 * systemd-nspawn's --network-veth= switch should now result in
16442 stable MAC addresses for both the outer and the inner side
16445 * systemd-nspawn gained a new --volatile= switch for running
16446 container instances with /etc or /var unpopulated.
16448 * The kdbus client code has been updated to use the new Linux
16449 3.17 memfd subsystem instead of the old kdbus-specific one.
16451 * systemd-networkd's DHCP client and server now support
16452 FORCERENEW. There are also new configuration options to
16453 configure the vendor client identifier and broadcast mode
16456 * systemd will no longer inform the kernel about the current
16457 timezone, as this is necessarily incorrect and racy as the
16458 kernel has no understanding of DST and similar
16459 concepts. This hence means FAT timestamps will be always
16460 considered UTC, similar to what Android is already
16461 doing. Also, when the RTC is configured to the local time
16462 (rather than UTC) systemd will never synchronize back to it,
16463 as this might confuse Windows at a later boot.
16465 * systemd-analyze gained a new command "verify" for offline
16466 validation of unit files.
16468 * systemd-networkd gained support for a couple of additional
16469 settings for bonding networking setups. Also, the metric for
16470 statically configured routes may now be configured. For
16471 network interfaces where this is appropriate the peer IP
16472 address may now be configured.
16474 * systemd-networkd's DHCP client will no longer request
16475 broadcasting by default, as this tripped up some networks.
16476 For hardware where broadcast is required the feature should
16477 be switched back on using RequestBroadcast=yes.
16479 * systemd-networkd will now set up IPv4LL addresses (when
16480 enabled) even if DHCP is configured successfully.
16482 * udev will now default to respect network device names given
16483 by the kernel when the kernel indicates that these are
16484 predictable. This behavior can be tweaked by changing
16485 NamePolicy= in the relevant .link file.
16487 * A new library systemd-terminal has been added that
16488 implements full TTY stream parsing and rendering. This
16489 library is supposed to be used later on for implementing a
16490 full userspace VT subsystem, replacing the current kernel
16493 * A new tool systemd-journal-upload has been added to push
16494 journal data to a remote system running
16495 systemd-journal-remote.
16497 * journald will no longer forward all local data to another
16498 running syslog daemon. This change has been made because
16499 rsyslog (which appears to be the most commonly used syslog
16500 implementation these days) no longer makes use of this, and
16501 instead pulls the data out of the journal on its own. Since
16502 forwarding the messages to a non-existent syslog server is
16503 more expensive than we assumed we have now turned this
16504 off. If you run a syslog server that is not a recent rsyslog
16505 version, you have to turn this option on again
16506 (ForwardToSyslog= in journald.conf).
16508 * journald now optionally supports the LZ4 compressor for
16509 larger journal fields. This compressor should perform much
16510 better than XZ which was the previous default.
16512 * machinectl now shows the IP addresses of local containers,
16513 if it knows them, plus the interface name of the container.
16515 * A new tool "systemd-escape" has been added that makes it
16516 easy to escape strings to build unit names and similar.
16518 * sd_notify() messages may now include a new ERRNO= field
16519 which is parsed and collected by systemd and shown among the
16520 "systemctl status" output for a service.
16522 * A new component "systemd-firstboot" has been added that
16523 queries the most basic systemd information (timezone,
16524 hostname, root password) interactively on first
16525 boot. Alternatively it may also be used to provision these
16526 things offline on OS images installed into directories.
16528 * The default sysctl.d/ snippets will now set
16530 net.ipv4.conf.default.promote_secondaries=1
16532 This has the benefit of no flushing secondary IP addresses
16533 when primary addresses are removed.
16535 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
16536 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
16537 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
16538 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
16539 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
16540 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
16541 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
16542 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
16543 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
16544 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
16545 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
16546 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
16547 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
16548 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
16549 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
16551 — Berlin, 2014-08-19
16555 * A new tool systemd-sysusers has been added. This tool
16556 creates system users and groups in /etc/passwd and
16557 /etc/group, based on static declarative system user/group
16558 definitions in /usr/lib/sysusers.d/. This is useful to
16559 enable factory resets and volatile systems that boot up with
16560 an empty /etc directory, and thus need system users and
16561 groups created during early boot. systemd now also ships
16562 with two default sysusers.d/ files for the most basic
16563 users and groups systemd and the core operating system
16566 * A new tmpfiles snippet has been added that rebuilds the
16567 essential files in /etc on boot, should they be missing.
16569 * A directive for ensuring automatic clean-up of
16570 /var/cache/man/ has been removed from the default
16571 configuration. This line should now be shipped by the man
16572 implementation. The necessary change has been made to the
16573 man-db implementation. Note that you need to update your man
16574 implementation to one that ships this line, otherwise no
16575 automatic clean-up of /var/cache/man will take place.
16577 * A new condition ConditionNeedsUpdate= has been added that
16578 may conditionalize services to only run when /etc or /var
16579 are "older" than the vendor operating system resources in
16580 /usr. This is useful for reconstructing or updating /etc
16581 after an offline update of /usr or a factory reset, on the
16582 next reboot. Services that want to run once after such an
16583 update or reset should use this condition and order
16584 themselves before the new systemd-update-done.service, which
16585 will mark the two directories as fully updated. A number of
16586 service files have been added making use of this, to rebuild
16587 the udev hardware database, the journald message catalog and
16588 dynamic loader cache (ldconfig). The systemd-sysusers tool
16589 described above also makes use of this now. With this in
16590 place it is now possible to start up a minimal operating
16591 system with /etc empty cleanly. For more information on the
16592 concepts involved see this recent blog story:
16594 https://0pointer.de/blog/projects/stateless.html
16596 * A new system group "input" has been introduced, and all
16597 input device nodes get this group assigned. This is useful
16598 for system-level software to get access to input devices. It
16599 complements what is already done for "audio" and "video".
16601 * systemd-networkd learnt minimal DHCPv4 server support in
16602 addition to the existing DHCPv4 client support. It also
16603 learnt DHCPv6 client and IPv6 Router Solicitation client
16604 support. The DHCPv4 client gained support for static routes
16605 passed in from the server. Note that the [DHCPv4] section
16606 known in older systemd-networkd versions has been renamed to
16607 [DHCP] and is now also used by the DHCPv6 client. Existing
16608 .network files using settings of this section should be
16609 updated, though compatibility is maintained. Optionally, the
16610 client hostname may now be sent to the DHCP server.
16612 * networkd gained support for vxlan virtual networks as well
16613 as tun/tap and dummy devices.
16615 * networkd gained support for automatic allocation of address
16616 ranges for interfaces from a system-wide pool of
16617 addresses. This is useful for dynamically managing a large
16618 number of interfaces with a single network configuration
16619 file. In particular this is useful to easily assign
16620 appropriate IP addresses to the veth links of a large number
16621 of nspawn instances.
16623 * RPM macros for processing sysusers, sysctl and binfmt
16624 drop-in snippets at package installation time have been
16627 * The /etc/os-release file should now be placed in
16628 /usr/lib/os-release. The old location is automatically
16629 created as symlink. /usr/lib is the more appropriate
16630 location of this file, since it shall actually describe the
16631 vendor operating system shipped in /usr, and not the
16632 configuration stored in /etc.
16634 * .mount units gained a new boolean SloppyOptions= setting
16635 that maps to mount(8)'s -s option which enables permissive
16636 parsing of unknown mount options.
16638 * tmpfiles learnt a new "L+" directive which creates a symlink
16639 but (unlike "L") deletes a pre-existing file first, should
16640 it already exist and not already be the correct
16641 symlink. Similarly, "b+", "c+" and "p+" directives have been
16642 added as well, which create block and character devices, as
16643 well as fifos in the filesystem, possibly removing any
16644 pre-existing files of different types.
16646 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
16647 'argument' field (which so far specified the source to
16648 symlink/copy the files from) is now optional. If omitted the
16649 same file os copied from /usr/share/factory/ suffixed by the
16650 full destination path. This is useful for populating /etc
16651 with essential files, by copying them from vendor defaults
16652 shipped in /usr/share/factory/etc.
16654 * A new command "systemctl preset-all" has been added that
16655 applies the service preset settings to all installed unit
16656 files. A new switch --preset-mode= has been added that
16657 controls whether only enable or only disable operations
16660 * A new command "systemctl is-system-running" has been added
16661 that allows checking the overall state of the system, for
16662 example whether it is fully up and running.
16664 * When the system boots up with an empty /etc, the equivalent
16665 to "systemctl preset-all" is executed during early boot, to
16666 make sure all default services are enabled after a factory
16669 * systemd now contains a minimal preset file that enables the
16670 most basic services systemd ships by default.
16672 * Unit files' [Install] section gained a new DefaultInstance=
16673 field for defining the default instance to create if a
16674 template unit is enabled with no instance specified.
16676 * A new passive target cryptsetup-pre.target has been added
16677 that may be used by services that need to make they run and
16678 finish before the first LUKS cryptographic device is set up.
16680 * The /dev/loop-control and /dev/btrfs-control device nodes
16681 are now owned by the "disk" group by default, opening up
16682 access to this group.
16684 * systemd-coredump will now automatically generate a
16685 stack trace of all core dumps taking place on the system,
16686 based on elfutils' libdw library. This stack trace is logged
16689 * systemd-coredump may now optionally store coredumps directly
16690 on disk (in /var/lib/systemd/coredump, possibly compressed),
16691 instead of storing them unconditionally in the journal. This
16692 mode is the new default. A new configuration file
16693 /etc/systemd/coredump.conf has been added to configure this
16694 and other parameters of systemd-coredump.
16696 * coredumpctl gained a new "info" verb to show details about a
16697 specific coredump. A new switch "-1" has also been added
16698 that makes sure to only show information about the most
16699 recent entry instead of all entries. Also, as the tool is
16700 generally useful now the "systemd-" prefix of the binary
16701 name has been removed. Distributions that want to maintain
16702 compatibility with the old name should add a symlink from
16703 the old name to the new name.
16705 * journald's SplitMode= now defaults to "uid". This makes sure
16706 that unprivileged users can access their own coredumps with
16707 coredumpctl without restrictions.
16709 * New kernel command line options "systemd.wants=" (for
16710 pulling an additional unit during boot), "systemd.mask="
16711 (for masking a specific unit for the boot), and
16712 "systemd.debug-shell" (for enabling the debug shell on tty9)
16713 have been added. This is implemented in the new generator
16714 "systemd-debug-generator".
16716 * systemd-nspawn will now by default filter a couple of
16717 syscalls for containers, among them those required for
16718 kernel module loading, direct x86 IO port access, swap
16719 management, and kexec. Most importantly though
16720 open_by_handle_at() is now prohibited for containers,
16721 closing a hole similar to a recently discussed vulnerability
16722 in docker regarding access to files on file hierarchies the
16723 container should normally not have access to. Note that, for
16724 nspawn, we generally make no security claims anyway (and
16725 this is explicitly documented in the man page), so this is
16726 just a fix for one of the most obvious problems.
16728 * A new man page file-hierarchy(7) has been added that
16729 contains a minimized, modernized version of the file system
16730 layout systemd expects, similar in style to the FHS
16731 specification or hier(5). A new tool systemd-path(1) has
16732 been added to query many of these paths for the local
16735 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
16736 longer done. Since the directory now has a per-user size
16737 limit, and is cleaned on logout this appears unnecessary,
16738 in particular since this now brings the lifecycle of this
16739 directory closer in line with how IPC objects are handled.
16741 * systemd.pc now exports a number of additional directories,
16742 including $libdir (which is useful to identify the library
16743 path for the primary architecture of the system), and a
16744 couple of drop-in directories.
16746 * udev's predictable network interface names now use the dev_port
16747 sysfs attribute, introduced in linux 3.15 instead of dev_id to
16748 distinguish between ports of the same PCI function. dev_id should
16749 only be used for ports using the same HW address, hence the need
16752 * machined has been updated to export the OS version of a
16753 container (read from /etc/os-release and
16754 /usr/lib/os-release) on the bus. This is now shown in
16755 "machinectl status" for a machine.
16757 * A new service setting RestartForceExitStatus= has been
16758 added. If configured to a set of exit signals or process
16759 return values, the service will be restarted when the main
16760 daemon process exits with any of them, regardless of the
16763 * systemctl's -H switch for connecting to remote systemd
16764 machines has been extended so that it may be used to
16765 directly connect to a specific container on the
16766 host. "systemctl -H root@foobar:waldi" will now connect as
16767 user "root" to host "foobar", and then proceed directly to
16768 the container named "waldi". Note that currently you have to
16769 authenticate as user "root" for this to work, as entering
16770 containers is a privileged operation.
16772 Contributions from: Andreas Henriksson, Benjamin Steinwender,
16773 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
16774 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
16775 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
16776 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
16777 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
16778 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
16779 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
16780 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
16781 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
16782 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
16783 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
16785 — Berlin, 2014-07-03
16789 * As an experimental feature, udev now tries to lock the
16790 disk device node (flock(LOCK_SH|LOCK_NB)) while it
16791 executes events for the disk or any of its partitions.
16792 Applications like partitioning programs can lock the
16793 disk device node (flock(LOCK_EX)) and claim temporary
16794 device ownership that way; udev will entirely skip all event
16795 handling for this disk and its partitions. If the disk
16796 was opened for writing, the close will trigger a partition
16797 table rescan in udev's "watch" facility, and if needed
16798 synthesize "change" events for the disk and all its partitions.
16799 This is now unconditionally enabled, and if it turns out to
16800 cause major problems, we might turn it on only for specific
16801 devices, or might need to disable it entirely. Device Mapper
16802 devices are excluded from this logic.
16804 * We temporarily dropped the "-l" switch for fsck invocations,
16805 since they collide with the flock() logic above. util-linux
16806 upstream has been changed already to avoid this conflict,
16807 and we will re-add "-l" as soon as util-linux with this
16808 change has been released.
16810 * The dependency on libattr has been removed. Since a long
16811 time, the extended attribute calls have moved to glibc, and
16812 libattr is thus unnecessary.
16814 * Virtualization detection works without privileges now. This
16815 means the systemd-detect-virt binary no longer requires
16816 CAP_SYS_PTRACE file capabilities, and our daemons can run
16817 with fewer privileges.
16819 * systemd-networkd now runs under its own "systemd-network"
16820 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
16821 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
16822 loses the ability to write to files owned by root this way.
16824 * Similarly, systemd-resolved now runs under its own
16825 "systemd-resolve" user with no capabilities remaining.
16827 * Similarly, systemd-bus-proxyd now runs under its own
16828 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
16830 * systemd-networkd gained support for setting up "veth"
16831 virtual Ethernet devices for container connectivity, as well
16832 as GRE and VTI tunnels.
16834 * systemd-networkd will no longer automatically attempt to
16835 manually load kernel modules necessary for certain tunnel
16836 transports. Instead, it is assumed the kernel loads them
16837 automatically when required. This only works correctly on
16838 very new kernels. On older kernels, please consider adding
16839 the kernel modules to /etc/modules-load.d/ as a work-around.
16841 * The resolv.conf file systemd-resolved generates has been
16842 moved to /run/systemd/resolve/. If you have a symlink from
16843 /etc/resolv.conf, it might be necessary to correct it.
16845 * Two new service settings, ProtectHome= and ProtectSystem=,
16846 have been added. When enabled, they will make the user data
16847 (such as /home) inaccessible or read-only and the system
16848 (such as /usr) read-only, for specific services. This allows
16849 very lightweight per-service sandboxing to avoid
16850 modifications of user data or system files from
16851 services. These two new switches have been enabled for all
16852 of systemd's long-running services, where appropriate.
16854 * Socket units gained new SocketUser= and SocketGroup=
16855 settings to set the owner user and group of AF_UNIX sockets
16856 and FIFOs in the file system.
16858 * Socket units gained a new RemoveOnStop= setting. If enabled,
16859 all FIFOS and sockets in the file system will be removed
16860 when the specific socket unit is stopped.
16862 * Socket units gained a new Symlinks= setting. It takes a list
16863 of symlinks to create to file system sockets or FIFOs
16864 created by the specific Unix sockets. This is useful to
16865 manage symlinks to socket nodes with the same lifecycle as
16868 * The /dev/log socket and /dev/initctl FIFO have been moved to
16869 /run, and have been replaced by symlinks. This allows
16870 connecting to these facilities even if PrivateDevices=yes is
16871 used for a service (which makes /dev/log itself unavailable,
16872 but /run is left). This also has the benefit of ensuring
16873 that /dev only contains device nodes, directories and
16874 symlinks, and nothing else.
16876 * sd-daemon gained two new calls sd_pid_notify() and
16877 sd_pid_notifyf(). They are similar to sd_notify() and
16878 sd_notifyf(), but allow overriding of the source PID of
16879 notification messages if permissions permit this. This is
16880 useful to send notify messages on behalf of a different
16881 process (for example, the parent process). The
16882 systemd-notify tool has been updated to make use of this
16883 when sending messages (so that notification messages now
16884 originate from the shell script invoking systemd-notify and
16885 not the systemd-notify process itself. This should minimize
16886 a race where systemd fails to associate notification
16887 messages to services when the originating process already
16890 * A new "on-abnormal" setting for Restart= has been added. If
16891 set, it will result in automatic restarts on all "abnormal"
16892 reasons for a process to exit, which includes unclean
16893 signals, core dumps, timeouts and watchdog timeouts, but
16894 does not include clean and unclean exit codes or clean
16895 signals. Restart=on-abnormal is an alternative for
16896 Restart=on-failure for services that shall be able to
16897 terminate and avoid restarts on certain errors, by
16898 indicating so with an unclean exit code. Restart=on-failure
16899 or Restart=on-abnormal is now the recommended setting for
16900 all long-running services.
16902 * If the InaccessibleDirectories= service setting points to a
16903 mount point (or if there are any submounts contained within
16904 it), it is now attempted to completely unmount it, to make
16905 the file systems truly unavailable for the respective
16908 * The ReadOnlyDirectories= service setting and
16909 systemd-nspawn's --read-only parameter are now recursively
16910 applied to all submounts, too.
16912 * Mount units may now be created transiently via the bus APIs.
16914 * The support for SysV and LSB init scripts has been removed
16915 from the systemd daemon itself. Instead, it is now
16916 implemented as a generator that creates native systemd units
16917 from these scripts when needed. This enables us to remove a
16918 substantial amount of legacy code from PID 1, following the
16919 fact that many distributions only ship a very small number
16920 of LSB/SysV init scripts nowadays.
16922 * Privileged Xen (dom0) domains are not considered
16923 virtualization anymore by the virtualization detection
16924 logic. After all, they generally have unrestricted access to
16925 the hardware and usually are used to manage the unprivileged
16928 * systemd-tmpfiles gained a new "C" line type, for copying
16929 files or entire directories.
16931 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
16932 lines. So far, they have been non-globbing versions of the
16933 latter, and have thus been redundant. In future, it is
16934 recommended to only use "z". "m" has hence been removed
16935 from the documentation, even though it stays supported.
16937 * A tmpfiles snippet to recreate the most basic structure in
16938 /var has been added. This is enough to create the /var/run →
16939 /run symlink and create a couple of structural
16940 directories. This allows systems to boot up with an empty or
16941 volatile /var. Of course, while with this change, the core OS
16942 now is capable with dealing with a volatile /var, not all
16943 user services are ready for it. However, we hope that sooner
16944 or later, many service daemons will be changed upstream so
16945 that they are able to automatically create their necessary
16946 directories in /var at boot, should they be missing. This is
16947 the first step to allow state-less systems that only require
16948 the vendor image for /usr to boot.
16950 * systemd-nspawn has gained a new --tmpfs= switch to mount an
16951 empty tmpfs instance to a specific directory. This is
16952 particularly useful for making use of the automatic
16953 reconstruction of /var (see above), by passing --tmpfs=/var.
16955 * Access modes specified in tmpfiles snippets may now be
16956 prefixed with "~", which indicates that they shall be masked
16957 by whether the existing file or directory is currently
16958 writable, readable or executable at all. Also, if specified,
16959 the sgid/suid/sticky bits will be masked for all
16962 * A new passive target unit "network-pre.target" has been
16963 added which is useful for services that shall run before any
16964 network is configured, for example firewall scripts.
16966 * The "floppy" group that previously owned the /dev/fd*
16967 devices is no longer used. The "disk" group is now used
16968 instead. Distributions should probably deprecate usage of
16971 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
16972 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
16973 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
16974 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
16975 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
16976 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
16979 — Berlin, 2014-06-11
16983 * A new "systemd-timesyncd" daemon has been added for
16984 synchronizing the system clock across the network. It
16985 implements an SNTP client. In contrast to NTP
16986 implementations such as chrony or the NTP reference server,
16987 this only implements a client side, and does not bother with
16988 the full NTP complexity, focusing only on querying time from
16989 one remote server and synchronizing the local clock to
16990 it. Unless you intend to serve NTP to networked clients or
16991 want to connect to local hardware clocks, this simple NTP
16992 client should be more than appropriate for most
16993 installations. The daemon runs with minimal privileges, and
16994 has been hooked up with networkd to only operate when
16995 network connectivity is available. The daemon saves the
16996 current clock to disk every time a new NTP sync has been
16997 acquired, and uses this to possibly correct the system clock
16998 early at bootup, in order to accommodate for systems that
16999 lack an RTC such as the Raspberry Pi and embedded devices,
17000 and to make sure that time monotonically progresses on these
17001 systems, even if it is not always correct. To make use of
17002 this daemon, a new system user and group "systemd-timesync"
17003 needs to be created on installation of systemd.
17005 * The queue "seqnum" interface of libudev has been disabled, as
17006 it was generally incompatible with device namespacing as
17007 sequence numbers of devices go "missing" if the devices are
17008 part of a different namespace.
17010 * "systemctl list-timers" and "systemctl list-sockets" gained
17011 a --recursive switch for showing units of these types also
17012 for all local containers, similar in style to the already
17013 supported --recursive switch for "systemctl list-units".
17015 * A new RebootArgument= setting has been added for service
17016 units, which may be used to specify a kernel reboot argument
17017 to use when triggering reboots with StartLimitAction=.
17019 * A new FailureAction= setting has been added for service
17020 units which may be used to specify an operation to trigger
17021 when a service fails. This works similarly to
17022 StartLimitAction=, but unlike it, controls what is done
17023 immediately rather than only after several attempts to
17024 restart the service in question.
17026 * hostnamed got updated to also expose the kernel name,
17027 release, and version on the bus. This is useful for
17028 executing commands like hostnamectl with the -H switch.
17029 systemd-analyze makes use of this to properly display
17030 details when running non-locally.
17032 * The bootchart tool can now show cgroup information in the
17033 graphs it generates.
17035 * The CFS CPU quota cgroup attribute is now exposed for
17036 services. The new CPUQuota= switch has been added for this
17037 which takes a percentage value. Setting this will have the
17038 result that a service may never get more CPU time than the
17039 specified percentage, even if the machine is otherwise idle.
17041 * systemd-networkd learned IPIP and SIT tunnel support.
17043 * LSB init scripts exposing a dependency on $network will now
17044 get a dependency on network-online.target rather than simply
17045 network.target. This should bring LSB handling closer to
17046 what it was on SysV systems.
17048 * A new fsck.repair= kernel option has been added to control
17049 how fsck shall deal with unclean file systems at boot.
17051 * The (.ini) configuration file parser will now silently ignore
17052 sections whose names begin with "X-". This may be used to maintain
17053 application-specific extension sections in unit files.
17055 * machined gained a new API to query the IP addresses of
17056 registered containers. "machinectl status" has been updated
17057 to show these addresses in its output.
17059 * A new call sd_uid_get_display() has been added to the
17060 sd-login APIs for querying the "primary" session of a
17061 user. The "primary" session of the user is elected from the
17062 user's sessions and generally a graphical session is
17063 preferred over a text one.
17065 * A minimal systemd-resolved daemon has been added. It
17066 currently simply acts as a companion to systemd-networkd and
17067 manages resolv.conf based on per-interface DNS
17068 configuration, possibly supplied via DHCP. In the long run
17069 we hope to extend this into a local DNSSEC enabled DNS and
17072 * The systemd-networkd-wait-online tool is now enabled by
17073 default. It will delay network-online.target until a network
17074 connection has been configured. The tool primarily integrates
17075 with networkd, but will also make a best effort to make sense
17076 of network configuration performed in some other way.
17078 * Two new service options StartupCPUShares= and
17079 StartupBlockIOWeight= have been added that work similarly to
17080 CPUShares= and BlockIOWeight= however only apply during
17081 system startup. This is useful to prioritize certain services
17082 differently during bootup than during normal runtime.
17084 * hostnamed has been changed to prefer the statically
17085 configured hostname in /etc/hostname (unless set to
17086 'localhost' or empty) over any dynamic one supplied by
17087 dhcp. With this change, the rules for picking the hostname
17088 match more closely the rules of other configuration settings
17089 where the local administrator's configuration in /etc always
17090 overrides any other settings.
17092 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
17093 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
17094 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
17095 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
17096 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
17097 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
17098 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
17099 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
17100 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
17101 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
17102 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
17103 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
17104 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
17105 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
17106 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
17107 Lindskog, WaLyong Cho, Will Woods, Zbigniew
17110 — Beijing, 2014-05-28
17114 * When restoring the screen brightness at boot, stay away from
17115 the darkest setting or from the lowest 5% of the available
17116 range, depending on which is the larger value of both. This
17117 should effectively protect the user from rebooting into a
17118 black screen, should the brightness have been set to minimum
17121 * sd-login gained a new sd_machine_get_class() call to
17122 determine the class ("vm" or "container") of a machine
17123 registered with machined.
17125 * sd-login gained new calls
17126 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
17127 to query the identity of the peer of a local AF_UNIX
17128 connection. They operate similarly to their sd_pid_get_xyz()
17131 * PID 1 will now maintain a system-wide system state engine
17132 with the states "starting", "running", "degraded",
17133 "maintenance", "stopping". These states are bound to system
17134 startup, normal runtime, runtime with at least one failed
17135 service, rescue/emergency mode and system shutdown. This
17136 state is shown in the "systemctl status" output when no unit
17137 name is passed. It is useful to determine system state, in
17138 particularly when doing so for many systems or containers at
17141 * A new command "list-machines" has been added to "systemctl"
17142 that lists all local OS containers and shows their system
17143 state (see above), if systemd runs inside of them.
17145 * systemctl gained a new "-r" switch to recursively enumerate
17146 units on all local containers, when used with the
17147 "list-unit" command (which is the default one that is
17148 executed when no parameters are specified).
17150 * The GPT automatic partition discovery logic will now honour
17151 two GPT partition flags: one may be set on a partition to
17152 cause it to be mounted read-only, and the other may be set
17153 on a partition to ignore it during automatic discovery.
17155 * Two new GPT type UUIDs have been added for automatic root
17156 partition discovery, for 32-bit and 64-bit ARM. This is not
17157 particularly useful for discovering the root directory on
17158 these architectures during bare-metal boots (since UEFI is
17159 not common there), but still very useful to allow booting of
17160 ARM disk images in nspawn with the -i option.
17162 * MAC addresses of interfaces created with nspawn's
17163 --network-interface= switch will now be generated from the
17164 machine name, and thus be stable between multiple invocations
17167 * logind will now automatically remove all IPC objects owned
17168 by a user if she or he fully logs out. This makes sure that
17169 users who are logged out cannot continue to consume IPC
17170 resources. This covers SysV memory, semaphores and message
17171 queues as well as POSIX shared memory and message
17172 queues. Traditionally, SysV and POSIX IPC had no lifecycle
17173 limits. With this functionality, that is corrected. This may
17174 be turned off by using the RemoveIPC= switch of logind.conf.
17176 * The systemd-machine-id-setup and tmpfiles tools gained a
17177 --root= switch to operate on a specific root directory,
17180 * journald can now forward logged messages to the TTYs of all
17181 logged in users ("wall"). This is the default for all
17182 emergency messages now.
17184 * A new tool systemd-journal-remote has been added to stream
17185 journal log messages across the network.
17187 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
17188 controller trees are mounted into it. Note that the
17189 directories mounted beneath it are not read-only. This is a
17190 security measure and is particularly useful because glibc
17191 actually includes a search logic to pick any tmpfs it can
17192 find to implement shm_open() if /dev/shm is not available
17193 (which it might very well be in namespaced setups).
17195 * machinectl gained a new "poweroff" command to cleanly power
17196 down a local OS container.
17198 * The PrivateDevices= unit file setting will now also drop the
17199 CAP_MKNOD capability from the capability bound set, and
17200 imply DevicePolicy=closed.
17202 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
17203 comprehensively on all long-running systemd services where
17204 this is appropriate.
17206 * systemd-udevd will now run in a disassociated mount
17207 namespace. To mount directories from udev rules, make sure to
17208 pull in mount units via SYSTEMD_WANTS properties.
17210 * The kdbus support gained support for uploading policy into
17211 the kernel. sd-bus gained support for creating "monitoring"
17212 connections that can eavesdrop into all bus communication
17213 for debugging purposes.
17215 * Timestamps may now be specified in seconds since the UNIX
17216 epoch Jan 1st, 1970 by specifying "@" followed by the value
17219 * Native tcpwrap support in systemd has been removed. tcpwrap
17220 is old code, not really maintained anymore and has serious
17221 shortcomings, and better options such as firewalls
17222 exist. For setups that require tcpwrap usage, please
17223 consider invoking your socket-activated service via tcpd,
17224 like on traditional inetd.
17226 * A new system.conf configuration option
17227 DefaultTimerAccuracySec= has been added that controls the
17228 default AccuracySec= setting of .timer units.
17230 * Timer units gained a new WakeSystem= switch. If enabled,
17231 timers configured this way will cause the system to resume
17232 from system suspend (if the system supports that, which most
17235 * Timer units gained a new Persistent= switch. If enabled,
17236 timers configured this way will save to disk when they have
17237 been last triggered. This information is then used on next
17238 reboot to possible execute overdue timer events, that
17239 could not take place because the system was powered off.
17240 This enables simple anacron-like behaviour for timer units.
17242 * systemctl's "list-timers" will now also list the time a
17243 timer unit was last triggered in addition to the next time
17244 it will be triggered.
17246 * systemd-networkd will now assign predictable IPv4LL
17247 addresses to its local interfaces.
17249 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
17250 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
17251 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
17252 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
17253 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
17254 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
17255 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
17256 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
17259 — Berlin, 2014-03-25
17263 * A new unit file setting RestrictAddressFamilies= has been
17264 added to restrict which socket address families unit
17265 processes gain access to. This takes address family names
17266 like "AF_INET" or "AF_UNIX", and is useful to minimize the
17267 attack surface of services via exotic protocol stacks. This
17268 is built on seccomp system call filters.
17270 * Two new unit file settings RuntimeDirectory= and
17271 RuntimeDirectoryMode= have been added that may be used to
17272 manage a per-daemon runtime directories below /run. This is
17273 an alternative for setting up directory permissions with
17274 tmpfiles snippets, and has the advantage that the runtime
17275 directory's lifetime is bound to the daemon runtime and that
17276 the daemon starts up with an empty directory each time. This
17277 is particularly useful when writing services that drop
17278 privileges using the User= or Group= setting.
17280 * The DeviceAllow= unit setting now supports globbing for
17281 matching against device group names.
17283 * The systemd configuration file system.conf gained new
17284 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
17285 DefaultMemoryAccounting= to globally turn on/off accounting
17286 for specific resources (cgroups) for all units. These
17287 settings may still be overridden individually in each unit
17290 * systemd-gpt-auto-generator is now able to discover /srv and
17291 root partitions in addition to /home and swap partitions. It
17292 also supports LUKS-encrypted partitions now. With this in
17293 place, automatic discovery of partitions to mount following
17294 the Discoverable Partitions Specification
17295 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
17296 is now a lot more complete. This allows booting without
17297 /etc/fstab and without root= on the kernel command line on
17298 systems prepared appropriately.
17300 * systemd-nspawn gained a new --image= switch which allows
17301 booting up disk images and Linux installations on any block
17302 device that follow the Discoverable Partitions Specification
17303 (see above). This means that installations made with
17304 appropriately updated installers may now be started and
17305 deployed using container managers, completely
17306 unmodified. (We hope that libvirt-lxc will add support for
17307 this feature soon, too.)
17309 * systemd-nspawn gained a new --network-macvlan= setting to
17310 set up a private macvlan interface for the
17311 container. Similarly, systemd-networkd gained a new
17312 Kind=macvlan setting in .netdev files.
17314 * systemd-networkd now supports configuring local addresses
17317 * A new tool systemd-network-wait-online has been added to
17318 synchronously wait for network connectivity using
17321 * The sd-bus.h bus API gained a new sd_bus_track object for
17322 tracking the lifecycle of bus peers. Note that sd-bus.h is
17323 still not a public API though (unless you specify
17324 --enable-kdbus on the configure command line, which however
17325 voids your warranty and you get no API stability guarantee).
17327 * The $XDG_RUNTIME_DIR runtime directories for each user are
17328 now individual tmpfs instances, which has the benefit of
17329 introducing separate pools for each user, with individual
17330 size limits, and thus making sure that unprivileged clients
17331 can no longer negatively impact the system or other users by
17332 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
17333 RuntimeDirectorySize= has been introduced that allows
17334 controlling the default size limit for all users. It
17335 defaults to 10% of the available physical memory. This is no
17336 replacement for quotas on tmpfs though (which the kernel
17337 still does not support), as /dev/shm and /tmp are still
17338 shared resources used by both the system and unprivileged
17341 * logind will now automatically turn off automatic suspending
17342 on laptop lid close when more than one display is
17343 connected. This was previously expected to be implemented
17344 individually in desktop environments (such as GNOME),
17345 however has been added to logind now, in order to fix a
17346 boot-time race where a desktop environment might not have
17347 been started yet and thus not been able to take an inhibitor
17348 lock at the time where logind already suspends the system
17349 due to a closed lid.
17351 * logind will now wait at least 30s after each system
17352 suspend/resume cycle, and 3min after system boot before
17353 suspending the system due to a closed laptop lid. This
17354 should give USB docking stations and similar enough time to
17355 be probed and configured after system resume and boot in
17356 order to then act as suspend blocker.
17358 * systemd-run gained a new --property= setting which allows
17359 initialization of resource control properties (and others)
17360 for the created scope or service unit. Example: "systemd-run
17361 --property=BlockIOWeight=10 updatedb" may be used to run
17362 updatedb at a low block IO scheduling weight.
17364 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
17365 now also work in --scope mode.
17367 * When systemd is compiled with kdbus support, basic support
17368 for enforced policies is now in place. (Note that enabling
17369 kdbus still voids your warranty and no API compatibility
17370 promises are made.)
17372 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
17373 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
17374 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
17375 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
17376 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
17377 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
17378 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
17379 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
17380 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
17381 Zbigniew Jędrzejewski-Szmek
17383 — Berlin, 2014-03-12
17387 * systemd will now relabel /dev after loading the SMACK policy
17388 according to SMACK rules.
17390 * A new unit file option AppArmorProfile= has been added to
17391 set the AppArmor profile for the processes of a unit.
17393 * A new condition check ConditionArchitecture= has been added
17394 to conditionalize units based on the system architecture, as
17395 reported by uname()'s "machine" field.
17397 * systemd-networkd now supports matching on the system
17398 virtualization, architecture, kernel command line, hostname
17401 * logind is now a lot more aggressive when suspending the
17402 machine due to a closed laptop lid. Instead of acting only
17403 on the lid close action, it will continuously watch the lid
17404 status and act on it. This is useful for laptops where the
17405 power button is on the outside of the chassis so that it can
17406 be reached without opening the lid (such as the Lenovo
17407 Yoga). On those machines, logind will now immediately
17408 re-suspend the machine if the power button has been
17409 accidentally pressed while the laptop was suspended and in a
17410 backpack or similar.
17412 * logind will now watch SW_DOCK switches and inhibit reaction
17413 to the lid switch if it is pressed. This means that logind
17414 will not suspend the machine anymore if the lid is closed
17415 and the system is docked, if the laptop supports SW_DOCK
17416 notifications via the input layer. Note that ACPI docking
17417 stations do not generate this currently. Also note that this
17418 logic is usually not fully sufficient and Desktop
17419 Environments should take a lid switch inhibitor lock when an
17420 external display is connected, as systemd will not watch
17423 * nspawn will now make use of the devices cgroup controller by
17424 default, and only permit creation of and access to the usual
17425 API device nodes like /dev/null or /dev/random, as well as
17426 access to (but not creation of) the pty devices.
17428 * We will now ship a default .network file for
17429 systemd-networkd that automatically configures DHCP for
17430 network interfaces created by nspawn's --network-veth or
17431 --network-bridge= switches.
17433 * systemd will now understand the usual M, K, G, T suffixes
17434 according to SI conventions (i.e. to the base 1000) when
17435 referring to throughput and hardware metrics. It will stay
17436 with IEC conventions (i.e. to the base 1024) for software
17437 metrics, according to what is customary according to
17438 Wikipedia. We explicitly document which base applies for
17439 each configuration option.
17441 * The DeviceAllow= setting in unit files now supports a syntax to
17442 allow-list an entire group of devices node majors at once, based on
17443 the /proc/devices listing. For example, with the string "char-pts",
17444 it is now possible to allow-list all current and future pseudo-TTYs
17447 * sd-event learned a new "post" event source. Event sources of
17448 this type are triggered by the dispatching of any event
17449 source of a type that is not "post". This is useful for
17450 implementing clean-up and check event sources that are
17451 triggered by other work being done in the program.
17453 * systemd-networkd is no longer statically enabled, but uses
17454 the usual [Install] sections so that it can be
17455 enabled/disabled using systemctl. It still is enabled by
17458 * When creating a veth interface pair with systemd-nspawn, the
17459 host side will now be prefixed with "vb-" if
17460 --network-bridge= is used, and with "ve-" if --network-veth
17461 is used. This way, it is easy to distinguish these cases on
17462 the host, for example to apply different configuration to
17463 them with systemd-networkd.
17465 * The compatibility libraries for libsystemd-journal.so,
17466 libsystem-id128.so, libsystemd-login.so and
17467 libsystemd-daemon.so do not make use of IFUNC
17468 anymore. Instead, we now build libsystemd.so multiple times
17469 under these alternative names. This means that the footprint
17470 is drastically increased, but given that these are
17471 transitional compatibility libraries, this should not matter
17472 much. This change has been made necessary to support the ARM
17473 platform for these compatibility libraries, as the ARM
17474 toolchain is not really at the same level as the toolchain
17475 for other architectures like x86 and does not support
17476 IFUNC. Please make sure to use --enable-compat-libs only
17477 during a transitional period!
17479 * The .include syntax has been deprecated and is not documented
17480 anymore. Drop-in files in .d directories should be used instead.
17482 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
17483 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
17484 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
17485 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
17486 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
17487 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
17488 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
17489 Zbigniew Jędrzejewski-Szmek
17491 — Berlin, 2014-02-24
17495 * A new component "systemd-networkd" has been added that can
17496 be used to configure local network interfaces statically or
17497 via DHCP. It is capable of bringing up bridges, VLANs, and
17498 bonding. Currently, no hook-ups for interactive network
17499 configuration are provided. Use this for your initrd,
17500 container, embedded, or server setup if you need a simple,
17501 yet powerful, network configuration solution. This
17502 configuration subsystem is quite nifty, as it allows wildcard
17503 hotplug matching in interfaces. For example, with a single
17504 configuration snippet, you can configure that all Ethernet
17505 interfaces showing up are automatically added to a bridge,
17506 or similar. It supports link-sensing and more.
17508 * A new tool "systemd-socket-proxyd" has been added which can
17509 act as a bidirectional proxy for TCP sockets. This is
17510 useful for adding socket activation support to services that
17511 do not actually support socket activation, including virtual
17512 machines and the like.
17514 * Add a new tool to save/restore rfkill state on
17517 * Save/restore state of keyboard backlights in addition to
17518 display backlights on shutdown/boot.
17520 * udev learned a new SECLABEL{} construct to label device
17521 nodes with a specific security label when they appear. For
17522 now, only SECLABEL{selinux} is supported, but the syntax is
17523 prepared for additional security frameworks.
17525 * udev gained a new scheme to configure link-level attributes
17526 from files in /etc/systemd/network/*.link. These files can
17527 match against MAC address, device path, driver name and type,
17528 and will apply attributes like the naming policy, link speed,
17529 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
17530 address assignment policy (randomized, …).
17532 * The configuration of network interface naming rules for
17533 "permanent interface names" has changed: a new NamePolicy=
17534 setting in the [Link] section of .link files determines the
17535 priority of possible naming schemes (onboard, slot, MAC,
17536 path). The default value of this setting is determined by
17537 /usr/lib/net/links/99-default.link. Old
17538 80-net-name-slot.rules udev configuration file has been
17539 removed, so local configuration overriding this file should
17540 be adapted to override 99-default.link instead.
17542 * When the User= switch is used in a unit file, also
17543 initialize $SHELL= based on the user database entry.
17545 * systemd no longer depends on libdbus. All communication is
17546 now done with sd-bus, systemd's low-level bus library
17549 * kdbus support has been added to PID 1 itself. When kdbus is
17550 enabled, this causes PID 1 to set up the system bus and
17551 enable support for a new ".busname" unit type that
17552 encapsulates bus name activation on kdbus. It works a little
17553 bit like ".socket" units, except for bus names. A new
17554 generator has been added that converts classic dbus1 service
17555 activation files automatically into native systemd .busname
17556 and .service units.
17558 * sd-bus: add a lightweight vtable implementation that allows
17559 defining objects on the bus with a simple static const
17560 vtable array of its methods, signals and properties.
17562 * systemd will not generate or install static dbus
17563 introspection data anymore to /usr/share/dbus-1/interfaces,
17564 as the precise format of these files is unclear, and
17565 nothing makes use of it.
17567 * A proxy daemon is now provided to proxy clients connecting
17568 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
17569 compatibility with classic D-Bus.
17571 * A bus driver implementation has been added that supports the
17572 classic D-Bus bus driver calls on kdbus, also for
17573 compatibility purposes.
17575 * A new API "sd-event.h" has been added that implements a
17576 minimal event loop API built around epoll. It provides a
17577 couple of features that direct epoll usage is lacking:
17578 prioritization of events, scales to large numbers of timer
17579 events, per-event timer slack (accuracy), system-wide
17580 coalescing of timer events, exit handlers, watchdog
17581 supervision support using systemd's sd_notify() API, child
17584 * A new API "sd-rntl.h" has been added that provides an API
17585 around the route netlink interface of the kernel, similar in
17586 style to "sd-bus.h".
17588 * A new API "sd-dhcp-client.h" has been added that provides a
17589 small DHCPv4 client-side implementation. This is used by
17590 "systemd-networkd".
17592 * There is a new kernel command line option
17593 "systemd.restore_state=0|1". When set to "0", none of the
17594 systemd tools will restore saved runtime state to hardware
17595 devices. More specifically, the rfkill and backlight states
17598 * The FsckPassNo= compatibility option in mount/service units
17599 has been removed. The fstab generator will now add the
17600 necessary dependencies automatically, and does not require
17601 PID1's support for that anymore.
17603 * journalctl gained a new switch, --list-boots, that lists
17604 recent boots with their times and boot IDs.
17606 * The various tools like systemctl, loginctl, timedatectl,
17607 busctl, systemd-run, … have gained a new switch "-M" to
17608 connect to a specific, local OS container (as direct
17609 connection, without requiring SSH). This works on any
17610 container that is registered with machined, such as those
17611 created by libvirt-lxc or nspawn.
17613 * systemd-run and systemd-analyze also gained support for "-H"
17614 to connect to remote hosts via SSH. This is particularly
17615 useful for systemd-run because it enables queuing of jobs
17616 onto remote systems.
17618 * machinectl gained a new command "login" to open a getty
17619 login in any local container. This works with any container
17620 that is registered with machined (such as those created by
17621 libvirt-lxc or nspawn), and which runs systemd inside.
17623 * machinectl gained a new "reboot" command that may be used to
17624 trigger a reboot on a specific container that is registered
17625 with machined. This works on any container that runs an init
17626 system of some kind.
17628 * systemctl gained a new "list-timers" command to print a nice
17629 listing of installed timer units with the times they elapse
17632 * Alternative reboot() parameters may now be specified on the
17633 "systemctl reboot" command line and are passed to the
17634 reboot() system call.
17636 * systemctl gained a new --job-mode= switch to configure the
17637 mode to queue a job with. This is a more generic version of
17638 --fail, --irreversible, and --ignore-dependencies, which are
17639 still available but not advertised anymore.
17641 * /etc/systemd/system.conf gained new settings to configure
17642 various default timeouts of units, as well as the default
17643 start limit interval and burst. These may still be overridden
17646 * PID1 will now export on the bus profile data of the security
17647 policy upload process (such as the SELinux policy upload to
17650 * journald: when forwarding logs to the console, include
17651 timestamps (following the setting in
17652 /sys/module/printk/parameters/time).
17654 * OnCalendar= in timer units now understands the special
17655 strings "yearly" and "annually". (Both are equivalent)
17657 * The accuracy of timer units is now configurable with the new
17658 AccuracySec= setting. It defaults to 1min.
17660 * A new dependency type JoinsNamespaceOf= has been added that
17661 allows running two services within the same /tmp and network
17662 namespace, if PrivateNetwork= or PrivateTmp= are used.
17664 * A new command "cat" has been added to systemctl. It outputs
17665 the original unit file of a unit, and concatenates the
17666 contents of additional "drop-in" unit file snippets, so that
17667 the full configuration is shown.
17669 * systemctl now supports globbing on the various "list-xyz"
17670 commands, like "list-units" or "list-sockets", as well as on
17671 those commands which take multiple unit names.
17673 * journalctl's --unit= switch gained support for globbing.
17675 * All systemd daemons now make use of the watchdog logic so
17676 that systemd automatically notices when they hang.
17678 * If the $container_ttys environment variable is set,
17679 getty-generator will automatically spawn a getty for each
17680 listed tty. This is useful for container managers to request
17681 login gettys to be spawned on as many ttys as needed.
17683 * %h, %s, %U specifier support is not available anymore when
17684 used in unit files for PID 1. This is because NSS calls are
17685 not safe from PID 1. They stay available for --user
17686 instances of systemd, and as special case for the root user.
17688 * loginctl gained a new "--no-legend" switch to turn off output
17689 of the legend text.
17691 * The "sd-login.h" API gained three new calls:
17692 sd_session_is_remote(), sd_session_get_remote_user(),
17693 sd_session_get_remote_host() to query information about
17696 * The udev hardware database now also carries vendor/product
17697 information of SDIO devices.
17699 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
17700 determine whether watchdog notifications are requested by
17701 the system manager.
17703 * Socket-activated per-connection services now include a
17704 short description of the connection parameters in the
17707 * tmpfiles gained a new "--boot" option. When this is not used,
17708 only lines where the command character is not suffixed with
17709 "!" are executed. When this option is specified, those
17710 options are executed too. This partitions tmpfiles
17711 directives into those that can be safely executed at any
17712 time, and those which should be run only at boot (for
17713 example, a line that creates /run/nologin).
17715 * A new API "sd-resolve.h" has been added which provides a simple
17716 asynchronous wrapper around glibc NSS hostname resolution
17717 calls, such as getaddrinfo(). In contrast to glibc's
17718 getaddrinfo_a(), it does not use signals. In contrast to most
17719 other asynchronous name resolution libraries, this one does
17720 not reimplement DNS, but reuses NSS, so that alternate
17721 hostname resolution systems continue to work, such as mDNS,
17722 LDAP, etc. This API is based on libasyncns, but it has been
17723 cleaned up for inclusion in systemd.
17725 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
17726 "sd-daemon.h" are no longer found in individual libraries
17727 libsystemd-journal.so, libsystemd-login.so,
17728 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
17729 merged them into a single library, libsystemd.so, which
17730 provides all symbols. The reason for this is cyclic
17731 dependencies, as these libraries tend to use each other's
17732 symbols. So far, we have managed to workaround that by linking
17733 a copy of a good part of our code into each of these
17734 libraries again and again, which, however, makes certain
17735 things hard to do, like sharing static variables. Also, it
17736 substantially increases footprint. With this change, there
17737 is only one library for the basic APIs systemd
17738 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
17739 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
17740 library as well, however are subject to the --enable-kdbus
17741 switch (see below). Note that "sd-dhcp-client.h" is not part
17742 of this library (this is because it only consumes, never
17743 provides, services of/to other APIs). To make the transition
17744 easy from the separate libraries to the unified one, we
17745 provide the --enable-compat-libs compile-time switch which
17746 will generate stub libraries that are compatible with the
17747 old ones but redirect all calls to the new one.
17749 * All of the kdbus logic and the new APIs "sd-bus.h",
17750 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
17751 and "sd-utf8.h" are compile-time optional via the
17752 "--enable-kdbus" switch, and they are not compiled in by
17753 default. To make use of kdbus, you have to explicitly enable
17754 the switch. Note however, that neither the kernel nor the
17755 userspace API for all of this is considered stable yet. We
17756 want to maintain the freedom to still change the APIs for
17757 now. By specifying this build-time switch, you acknowledge
17758 that you are aware of the instability of the current
17761 * Also, note that while kdbus is pretty much complete,
17762 it lacks one thing: proper policy support. This means you
17763 can build a fully working system with all features; however,
17764 it will be highly insecure. Policy support will be added in
17765 one of the next releases, at the same time that we will
17766 declare the APIs stable.
17768 * When the kernel command line argument "kdbus" is specified,
17769 systemd will automatically load the kdbus.ko kernel module. At
17770 this stage of development, it is only useful for testing kdbus
17771 and should not be used in production. Note: if "--enable-kdbus"
17772 is specified, and the kdbus.ko kernel module is available, and
17773 "kdbus" is added to the kernel command line, the entire system
17774 runs with kdbus instead of dbus-daemon, with the above mentioned
17775 problem of missing the system policy enforcement. Also a future
17776 version of kdbus.ko or a newer systemd will not be compatible with
17777 each other, and will unlikely be able to boot the machine if only
17778 one of them is updated.
17780 * systemctl gained a new "import-environment" command which
17781 uploads the caller's environment (or parts thereof) into the
17782 service manager so that it is inherited by services started
17783 by the manager. This is useful to upload variables like
17784 $DISPLAY into the user service manager.
17786 * A new PrivateDevices= switch has been added to service units
17787 which allows running a service with a namespaced /dev
17788 directory that does not contain any device nodes for
17789 physical devices. More specifically, it only includes devices
17790 such as /dev/null, /dev/urandom, and /dev/zero which are API
17793 * logind has been extended to support behaviour like VT
17794 switching on seats that do not support a VT. This makes
17795 multi-session available on seats that are not the first seat
17796 (seat0), and on systems where kernel support for VTs has
17797 been disabled at compile-time.
17799 * If a process holds a delay lock for system sleep or shutdown
17800 and fails to release it in time, we will now log its
17801 identity. This makes it easier to identify processes that
17802 cause slow suspends or power-offs.
17804 * When parsing /etc/crypttab, support for a new key-slot=
17805 option as supported by Debian is added. It allows indicating
17806 which LUKS slot to use on disk, speeding up key loading.
17808 * The sd_journal_sendv() API call has been checked and
17809 officially declared to be async-signal-safe so that it may
17810 be invoked from signal handlers for logging purposes.
17812 * Boot-time status output is now enabled automatically after a
17813 short timeout if boot does not progress, in order to give
17814 the user an indication what she or he is waiting for.
17816 * The boot-time output has been improved to show how much time
17817 remains until jobs expire.
17819 * The KillMode= switch in service units gained a new possible
17820 value "mixed". If set, and the unit is shut down, then the
17821 initial SIGTERM signal is sent only to the main daemon
17822 process, while the following SIGKILL signal is sent to
17823 all remaining processes of the service.
17825 * When a scope unit is registered, a new property "Controller"
17826 may be set. If set to a valid bus name, systemd will send a
17827 RequestStop() signal to this name when it would like to shut
17828 down the scope. This may be used to hook manager logic into
17829 the shutdown logic of scope units. Also, scope units may now
17830 be put in a special "abandoned" state, in which case the
17831 manager process which created them takes no further
17832 responsibilities for it.
17834 * When reading unit files, systemd will now verify
17835 the access mode of these files, and warn about certain
17836 suspicious combinations. This has been added to make it
17837 easier to track down packaging bugs where unit files are
17838 marked executable or world-writable.
17840 * systemd-nspawn gained a new "--setenv=" switch to set
17841 container-wide environment variables. The similar option in
17842 systemd-activate was renamed from "--environment=" to
17843 "--setenv=" for consistency.
17845 * systemd-nspawn has been updated to create a new kdbus domain
17846 for each container that is invoked, thus allowing each
17847 container to have its own set of system and user buses,
17848 independent of the host.
17850 * systemd-nspawn gained a new --drop-capability= switch to run
17851 the container with less capabilities than the default. Both
17852 --drop-capability= and --capability= now take the special
17853 string "all" for dropping or keeping all capabilities.
17855 * systemd-nspawn gained new switches for executing containers
17856 with specific SELinux labels set.
17858 * systemd-nspawn gained a new --quiet switch to not generate
17859 any additional output but the container's own console
17862 * systemd-nspawn gained a new --share-system switch to run a
17863 container without PID namespacing enabled.
17865 * systemd-nspawn gained a new --register= switch to control
17866 whether the container is registered with systemd-machined or
17867 not. This is useful for containers that do not run full
17868 OS images, but only specific apps.
17870 * systemd-nspawn gained a new --keep-unit which may be used
17871 when invoked as the only program from a service unit, and
17872 results in registration of the unit service itself in
17873 systemd-machined, instead of a newly opened scope unit.
17875 * systemd-nspawn gained a new --network-interface= switch for
17876 moving arbitrary interfaces to the container. The new
17877 --network-veth switch creates a virtual Ethernet connection
17878 between host and container. The new --network-bridge=
17879 switch then allows assigning the host side of this virtual
17880 Ethernet connection to a bridge device.
17882 * systemd-nspawn gained a new --personality= switch for
17883 setting the kernel personality for the container. This is
17884 useful when running a 32-bit container on a 64-bit host. A
17885 similar option Personality= is now also available for service
17888 * logind will now also track a "Desktop" identifier for each
17889 session which encodes the desktop environment of it. This is
17890 useful for desktop environments that want to identify
17891 multiple running sessions of itself easily.
17893 * A new SELinuxContext= setting for service units has been
17894 added that allows setting a specific SELinux execution
17895 context for a service.
17897 * Most systemd client tools will now honour $SYSTEMD_LESS for
17898 settings of the "less" pager. By default, these tools will
17899 override $LESS to allow certain operations to work, such as
17900 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
17901 influence this logic.
17903 * systemd's "seccomp" hook-up has been changed to make use of
17904 the libseccomp library instead of using its own
17905 implementation. This has benefits for portability among
17908 * For usage together with SystemCallFilter=, a new
17909 SystemCallErrorNumber= setting has been introduced that
17910 allows configuration of a system error number to be returned
17911 on filtered system calls, instead of immediately killing the
17912 process. Also, SystemCallArchitectures= has been added to
17913 limit access to system calls of a particular architecture
17914 (in order to turn off support for unused secondary
17915 architectures). There is also a global
17916 SystemCallArchitectures= setting in system.conf now to turn
17917 off support for non-native system calls system-wide.
17919 * systemd requires a kernel with a working name_to_handle_at(),
17920 please see the kernel config requirements in the README file.
17922 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
17923 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
17924 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
17925 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
17926 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
17927 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
17928 Elia Pinto, Florian Weimer, George McCollister, Goffredo
17929 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
17930 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
17931 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
17932 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
17933 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
17934 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
17935 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
17936 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
17937 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
17938 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
17939 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
17940 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
17941 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
17942 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
17943 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
17944 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
17945 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
17947 — Berlin, 2014-02-20
17951 * logind has gained support for facilitating privileged input
17952 and drm device access for unprivileged clients. This work is
17953 useful to allow Wayland display servers (and similar
17954 programs, such as kmscon) to run under the user's ID and
17955 access input and drm devices which are normally
17956 protected. When this is used (and the kernel is new enough)
17957 logind will "mute" IO on the file descriptors passed to
17958 Wayland as long as it is in the background and "unmute" it
17959 if it returns into the foreground. This allows secure
17960 session switching without allowing background sessions to
17961 eavesdrop on input and display data. This also introduces
17962 session switching support if VT support is turned off in the
17963 kernel, and on seats that are not seat0.
17965 * A new kernel command line option luks.options= is understood
17966 now which allows specifying LUKS options for usage for LUKS
17967 encrypted partitions specified with luks.uuid=.
17969 * tmpfiles.d(5) snippets may now use specifier expansion in
17970 path names. More specifically %m, %b, %H, %v, are now
17971 replaced by the local machine id, boot id, hostname, and
17972 kernel version number.
17974 * A new tmpfiles.d(5) command "m" has been introduced which
17975 may be used to change the owner/group/access mode of a file
17976 or directory if it exists, but do nothing if it does not.
17978 * This release removes high-level support for the
17979 MemorySoftLimit= cgroup setting. The underlying kernel
17980 cgroup attribute memory.soft_limit= is currently badly
17981 designed and likely to be removed from the kernel API in its
17982 current form, hence we should not expose it for now.
17984 * The memory.use_hierarchy cgroup attribute is now enabled for
17985 all cgroups systemd creates in the memory cgroup
17986 hierarchy. This option is likely to be come the built-in
17987 default in the kernel anyway, and the non-hierarchical mode
17988 never made much sense in the intrinsically hierarchical
17991 * A new field _SYSTEMD_SLICE= is logged along with all journal
17992 messages containing the slice a message was generated
17993 from. This is useful to allow easy per-customer filtering of
17994 logs among other things.
17996 * systemd-journald will no longer adjust the group of journal
17997 files it creates to the "systemd-journal" group. Instead we
17998 rely on the journal directory to be owned by the
17999 "systemd-journal" group, and its setgid bit set, so that the
18000 kernel file system layer will automatically enforce that
18001 journal files inherit this group assignment. The reason for
18002 this change is that we cannot allow NSS look-ups from
18003 journald which would be necessary to resolve
18004 "systemd-journal" to a numeric GID, because this might
18005 create deadlocks if NSS involves synchronous queries to
18006 other daemons (such as nscd, or sssd) which in turn are
18007 logging clients of journald and might block on it, which
18008 would then dead lock. A tmpfiles.d(5) snippet included in
18009 systemd will make sure the setgid bit and group are
18010 properly set on the journal directory if it exists on every
18011 boot. However, we recommend adjusting it manually after
18012 upgrades too (or from RPM scriptlets), so that the change is
18013 not delayed until next reboot.
18015 * Backlight and random seed files in /var/lib/ have moved into
18016 the /var/lib/systemd/ directory, in order to centralize all
18017 systemd generated files in one directory.
18019 * Boot time performance measurements (as displayed by
18020 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
18021 performance information if that's available to determine how
18022 much time BIOS and boot loader initialization required. With
18023 a sufficiently new BIOS you hence no longer need to boot
18024 with Gummiboot to get access to such information.
18026 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
18027 Cristian Rodríguez, Dave Reisner, David Herrmann, David
18028 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
18029 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
18030 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
18031 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
18032 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
18034 — Berlin, 2013-10-02
18038 * The Restart= option for services now understands a new
18039 on-watchdog setting, which will restart the service
18040 automatically if the service stops sending out watchdog keep
18041 alive messages (as configured with WatchdogSec=).
18043 * The getty generator (which is responsible for bringing up a
18044 getty on configured serial consoles) will no longer only
18045 start a getty on the primary kernel console but on all
18046 others, too. This makes the order in which console= is
18047 specified on the kernel command line less important.
18049 * libsystemd-logind gained a new sd_session_get_vt() call to
18050 retrieve the VT number of a session.
18052 * If the option "tries=0" is set for an entry of /etc/crypttab
18053 its passphrase is queried indefinitely instead of any
18054 maximum number of tries.
18056 * If a service with a configure PID file terminates its PID
18057 file will now be removed automatically if it still exists
18058 afterwards. This should put an end to stale PID files.
18060 * systemd-run will now also take relative binary path names
18061 for execution and no longer insists on absolute paths.
18063 * InaccessibleDirectories= and ReadOnlyDirectories= now take
18064 paths that are optionally prefixed with "-" to indicate that
18065 it should not be considered a failure if they do not exist.
18067 * journalctl -o (and similar commands) now understands a new
18068 output mode "short-precise", it is similar to "short" but
18069 shows timestamps with usec accuracy.
18071 * The option "discard" (as known from Debian) is now
18072 synonymous to "allow-discards" in /etc/crypttab. In fact,
18073 "discard" is preferred now (since it is easier to remember
18076 * Some licensing clean-ups were made, so that more code is now
18077 LGPL-2.1 licensed than before.
18079 * A minimal tool to save/restore the display backlight
18080 brightness across reboots has been added. It will store the
18081 backlight setting as late as possible at shutdown, and
18082 restore it as early as possible during reboot.
18084 * A logic to automatically discover and enable home and swap
18085 partitions on GPT disks has been added. With this in place
18086 /etc/fstab becomes optional for many setups as systemd can
18087 discover certain partitions located on the root disk
18088 automatically. Home partitions are recognized under their
18089 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
18090 partitions are recognized under their GPT type ID
18091 0657fd6da4ab43c484e50933c84b4f4f.
18093 * systemd will no longer pass any environment from the kernel
18094 or initrd to system services. If you want to set an
18095 environment for all services, do so via the kernel command
18096 line systemd.setenv= assignment.
18098 * The systemd-sysctl tool no longer natively reads the file
18099 /etc/sysctl.conf. If desired, the file should be symlinked
18100 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
18101 legacy support by a symlink rather than built-in code, it
18102 also makes the otherwise hidden order of application of the
18103 different files visible. (Note that this partly reverts to a
18104 pre-198 application order of sysctl knobs!)
18106 * The "systemctl set-log-level" and "systemctl dump" commands
18107 have been moved to systemd-analyze.
18109 * systemd-run learned the new --remain-after-exit switch,
18110 which causes the scope unit not to be cleaned up
18111 automatically after the process terminated.
18113 * tmpfiles learned a new --exclude-prefix= switch to exclude
18114 certain paths from operation.
18116 * journald will now automatically flush all messages to disk
18117 as soon as a message at the log level CRIT, ALERT or EMERG
18120 Contributions from: Andrew Cook, Brandon Philips, Christian
18121 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
18122 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
18123 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
18124 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
18125 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
18126 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
18127 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
18128 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
18129 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
18130 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
18131 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
18132 William Giokas, Zbigniew Jędrzejewski-Szmek
18134 — Berlin, 2013-09-13
18138 * The documentation has been updated to cover the various new
18139 concepts introduced with 205.
18141 * Unit files now understand the new %v specifier which
18142 resolves to the kernel version string as returned by "uname
18145 * systemctl now supports filtering the unit list output by
18146 load state, active state and sub state, using the new
18147 --state= parameter.
18149 * "systemctl status" will now show the results of the
18150 condition checks (like ConditionPathExists= and similar) of
18151 the last start attempts of the unit. They are also logged to
18154 * "journalctl -b" may now be used to look for boot output of a
18155 specific boot. Try "journalctl -b -1" for the previous boot,
18156 but the syntax is substantially more powerful.
18158 * "journalctl --show-cursor" has been added which prints the
18159 cursor string the last shown log line. This may then be used
18160 with the new "journalctl --after-cursor=" switch to continue
18161 browsing logs from that point on.
18163 * "journalctl --force" may now be used to force regeneration
18166 * Creation of "dead" device nodes has been moved from udev
18167 into kmod and tmpfiles. Previously, udev would read the kmod
18168 databases to pre-generate dead device nodes based on meta
18169 information contained in kernel modules, so that these would
18170 be auto-loaded on access rather then at boot. As this
18171 does not really have much to do with the exposing actual
18172 kernel devices to userspace this has always been slightly
18173 alien in the udev codebase. Following the new scheme kmod
18174 will now generate a runtime snippet for tmpfiles from the
18175 module meta information and it now is tmpfiles' job to the
18176 create the nodes. This also allows overriding access and
18177 other parameters for the nodes using the usual tmpfiles
18178 facilities. As side effect this allows us to remove the
18179 CAP_SYS_MKNOD capability bit from udevd entirely.
18181 * logind's device ACLs may now be applied to these "dead"
18182 devices nodes too, thus finally allowing managed access to
18183 devices such as /dev/snd/sequencer without loading the
18184 backing module right-away.
18186 * A new RPM macro has been added that may be used to apply
18187 tmpfiles configuration during package installation.
18189 * systemd-detect-virt and ConditionVirtualization= now can
18190 detect User-Mode-Linux machines (UML).
18192 * journald will now implicitly log the effective capabilities
18193 set of processes in the message metadata.
18195 * systemd-cryptsetup has gained support for TrueCrypt volumes.
18197 * The initrd interface has been simplified (more specifically,
18198 support for passing performance data via environment
18199 variables and fsck results via files in /run has been
18200 removed). These features were non-essential, and are
18201 nowadays available in a much nicer way by having systemd in
18202 the initrd serialize its state and have the hosts systemd
18203 deserialize it again.
18205 * The udev "keymap" data files and tools to apply keyboard
18206 specific mappings of scan to key codes, and force-release
18207 scan code lists have been entirely replaced by a udev
18208 "keyboard" builtin and a hwdb data file.
18210 * systemd will now honour the kernel's "quiet" command line
18211 argument also during late shutdown, resulting in a
18212 completely silent shutdown when used.
18214 * There's now an option to control the SO_REUSEPORT socket
18215 option in .socket units.
18217 * Instance units will now automatically get a per-template
18218 subslice of system.slice unless something else is explicitly
18219 configured. For example, instances of sshd@.service will now
18220 implicitly be placed in system-sshd.slice rather than
18221 system.slice as before.
18223 * Test coverage support may now be enabled at build time.
18225 Contributions from: Dave Reisner, Frederic Crozat, Harald
18226 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
18227 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
18228 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
18229 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
18230 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
18231 Giokas, Zbigniew Jędrzejewski-Szmek
18233 — Berlin, 2013-07-23
18237 * Two new unit types have been introduced:
18239 Scope units are very similar to service units, however, are
18240 created out of pre-existing processes — instead of PID 1
18241 forking off the processes. By using scope units it is
18242 possible for system services and applications to group their
18243 own child processes (worker processes) in a powerful way
18244 which then maybe used to organize them, or kill them
18245 together, or apply resource limits on them.
18247 Slice units may be used to partition system resources in an
18248 hierarchical fashion and then assign other units to them. By
18249 default there are now three slices: system.slice (for all
18250 system services), user.slice (for all user sessions),
18251 machine.slice (for VMs and containers).
18253 Slices and scopes have been introduced primarily in
18254 context of the work to move cgroup handling to a
18255 single-writer scheme, where only PID 1
18256 creates/removes/manages cgroups.
18258 * There's a new concept of "transient" units. In contrast to
18259 normal units these units are created via an API at runtime,
18260 not from configuration from disk. More specifically this
18261 means it is now possible to run arbitrary programs as
18262 independent services, with all execution parameters passed
18263 in via bus APIs rather than read from disk. Transient units
18264 make systemd substantially more dynamic then it ever was,
18265 and useful as a general batch manager.
18267 * logind has been updated to make use of scope and slice units
18268 for managing user sessions. As a user logs in he will get
18269 his own private slice unit, to which all sessions are added
18270 as scope units. We also added support for automatically
18271 adding an instance of user@.service for the user into the
18272 slice. Effectively logind will no longer create cgroup
18273 hierarchies on its own now, it will defer entirely to PID 1
18274 for this by means of scope, service and slice units. Since
18275 user sessions this way become entities managed by PID 1
18276 the output of "systemctl" is now a lot more comprehensive.
18278 * A new mini-daemon "systemd-machined" has been added which
18279 may be used by virtualization managers to register local
18280 VMs/containers. nspawn has been updated accordingly, and
18281 libvirt will be updated shortly. machined will collect a bit
18282 of meta information about the VMs/containers, and assign
18283 them their own scope unit (see above). The collected
18284 meta-data is then made available via the "machinectl" tool,
18285 and exposed in "ps" and similar tools. machined/machinectl
18286 is compile-time optional.
18288 * As discussed earlier, the low-level cgroup configuration
18289 options ControlGroup=, ControlGroupModify=,
18290 ControlGroupPersistent=, ControlGroupAttribute= have been
18291 removed. Please use high-level attribute settings instead as
18292 well as slice units.
18294 * A new bus call SetUnitProperties() has been added to alter
18295 various runtime parameters of a unit. This is primarily
18296 useful to alter cgroup parameters dynamically in a nice way,
18297 but will be extended later on to make more properties
18298 modifiable at runtime. systemctl gained a new set-properties
18299 command that wraps this call.
18301 * A new tool "systemd-run" has been added which can be used to
18302 run arbitrary command lines as transient services or scopes,
18303 while configuring a number of settings via the command
18304 line. This tool is currently very basic, however already
18305 very useful. We plan to extend this tool to even allow
18306 queuing of execution jobs with time triggers from the
18307 command line, similar in fashion to "at".
18309 * nspawn will now inform the user explicitly that kernels with
18310 audit enabled break containers, and suggest the user to turn
18313 * Support for detecting the IMA and AppArmor security
18314 frameworks with ConditionSecurity= has been added.
18316 * journalctl gained a new "-k" switch for showing only kernel
18317 messages, mimicking dmesg output; in addition to "--user"
18318 and "--system" switches for showing only user's own logs
18321 * systemd-delta can now show information about drop-in
18322 snippets extending unit files.
18324 * libsystemd-bus has been substantially updated but is still
18325 not available as public API.
18327 * systemd will now look for the "debug" argument on the kernel
18328 command line and enable debug logging, similar to what
18329 "systemd.log_level=debug" already did before.
18331 * "systemctl set-default", "systemctl get-default" has been
18332 added to configure the default.target symlink, which
18333 controls what to boot into by default.
18335 * "systemctl set-log-level" has been added as a convenient
18336 way to raise and lower systemd logging threshold.
18338 * "systemd-analyze plot" will now show the time the various
18339 generators needed for execution, as well as information
18340 about the unit file loading.
18342 * libsystemd-journal gained a new sd_journal_open_files() call
18343 for opening specific journal files. journactl also gained a
18344 new switch to expose this new functionality. Previously we
18345 only supported opening all files from a directory, or all
18346 files from the system, as opening individual files only is
18347 racy due to journal file rotation.
18349 * systemd gained the new DefaultEnvironment= setting in
18350 /etc/systemd/system.conf to set environment variables for
18353 * If a privileged process logs a journal message with the
18354 OBJECT_PID= field set, then journald will automatically
18355 augment this with additional OBJECT_UID=, OBJECT_GID=,
18356 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
18357 system services want to log events about specific client
18358 processes. journactl/systemctl has been updated to make use
18359 of this information if all log messages regarding a specific
18362 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
18363 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
18364 Reisner, David Coppa, David King, David Strauss, Eelco
18365 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
18366 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
18367 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
18368 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
18369 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
18370 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
18371 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
18372 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
18373 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
18374 Łukasz Stelmach, 장동준
18378 * The Python bindings gained some minimal support for the APIs
18379 exposed by libsystemd-logind.
18381 * ConditionSecurity= gained support for detecting SMACK. Since
18382 this condition already supports SELinux and AppArmor we only
18383 miss IMA for this. Patches welcome!
18385 Contributions from: Karol Lewandowski, Lennart Poettering,
18386 Zbigniew Jędrzejewski-Szmek
18390 * systemd-nspawn will now create /etc/resolv.conf if
18391 necessary, before bind-mounting the host's file onto it.
18393 * systemd-nspawn will now store meta information about a
18394 container on the container's cgroup as extended attribute
18395 fields, including the root directory.
18397 * The cgroup hierarchy has been reworked in many ways. All
18398 objects any of the components systemd creates in the cgroup
18399 tree are now suffixed. More specifically, user sessions are
18400 now placed in cgroups suffixed with ".session", users in
18401 cgroups suffixed with ".user", and nspawn containers in
18402 cgroups suffixed with ".nspawn". Furthermore, all cgroup
18403 names are now escaped in a simple scheme to avoid collision
18404 of userspace object names with kernel filenames. This work
18405 is preparation for making these objects relocatable in the
18406 cgroup tree, in order to allow easy resource partitioning of
18407 these objects without causing naming conflicts.
18409 * systemctl list-dependencies gained the new switches
18410 --plain, --reverse, --after and --before.
18412 * systemd-inhibit now shows the process name of processes that
18413 have taken an inhibitor lock.
18415 * nss-myhostname will now also resolve "localhost"
18416 implicitly. This makes /etc/hosts an optional file and
18417 nicely handles that on IPv6 ::1 maps to both "localhost" and
18418 the local hostname.
18420 * libsystemd-logind.so gained a new call
18421 sd_get_machine_names() to enumerate running containers and
18422 VMs (currently only supported by very new libvirt and
18423 nspawn). sd_login_monitor can now be used to watch
18424 VMs/containers coming and going.
18426 * .include is not allowed recursively anymore, and only in
18427 unit files. Usually it is better to use drop-in snippets in
18428 .d/*.conf anyway, as introduced with systemd 198.
18430 * systemd-analyze gained a new "critical-chain" command that
18431 determines the slowest chain of units run during system
18432 boot-up. It is very useful for tracking down where
18433 optimizing boot time is the most beneficial.
18435 * systemd will no longer allow manipulating service paths in
18436 the name=systemd:/system cgroup tree using ControlGroup= in
18437 units. (But is still fine with it in all other dirs.)
18439 * There's a new systemd-nspawn@.service service file that may
18440 be used to easily run nspawn containers as system
18441 services. With the container's root directory in
18442 /var/lib/container/foobar it is now sufficient to run
18443 "systemctl start systemd-nspawn@foobar.service" to boot it.
18445 * systemd-cgls gained a new parameter "--machine" to list only
18446 the processes within a certain container.
18448 * ConditionSecurity= now can check for "apparmor". We still
18449 are lacking checks for SMACK and IMA for this condition
18450 check though. Patches welcome!
18452 * A new configuration file /etc/systemd/sleep.conf has been
18453 added that may be used to configure which kernel operation
18454 systemd is supposed to execute when "suspend", "hibernate"
18455 or "hybrid-sleep" is requested. This makes the new kernel
18456 "freeze" state accessible to the user.
18458 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
18459 the passed argument if applicable.
18461 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
18462 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
18463 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
18464 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
18465 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
18466 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
18467 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
18472 * The output of 'systemctl list-jobs' got some polishing. The
18473 '--type=' argument may now be passed more than once. A new
18474 command 'systemctl list-sockets' has been added which shows
18475 a list of kernel sockets systemd is listening on with the
18476 socket units they belong to, plus the units these socket
18479 * The experimental libsystemd-bus library got substantial
18480 updates to work in conjunction with the (also experimental)
18481 kdbus kernel project. It works well enough to exchange
18482 messages with some sophistication. Note that kdbus is not
18483 ready yet, and the library is mostly an elaborate test case
18484 for now, and not installable.
18486 * systemd gained a new unit 'systemd-static-nodes.service'
18487 that generates static device nodes earlier during boot, and
18488 can run in conjunction with udev.
18490 * libsystemd-login gained a new call sd_pid_get_user_unit()
18491 to retrieve the user systemd unit a process is running
18492 in. This is useful for systems where systemd is used as
18495 * systemd-nspawn now places all containers in the new /machine
18496 top-level cgroup directory in the name=systemd
18497 hierarchy. libvirt will soon do the same, so that we get a
18498 uniform separation of /system, /user and /machine for system
18499 services, user processes and containers/virtual
18500 machines. This new cgroup hierarchy is also useful to stick
18501 stable names to specific container instances, which can be
18502 recognized later this way (this name may be controlled
18503 via systemd-nspawn's new -M switch). libsystemd-login also
18504 gained a new call sd_pid_get_machine_name() to retrieve the
18505 name of the container/VM a specific process belongs to.
18507 * bootchart can now store its data in the journal.
18509 * libsystemd-journal gained a new call
18510 sd_journal_add_conjunction() for AND expressions to the
18511 matching logic. This can be used to express more complex
18512 logical expressions.
18514 * journactl can now take multiple --unit= and --user-unit=
18517 * The cryptsetup logic now understands the "luks.key=" kernel
18518 command line switch for specifying a file to read the
18519 decryption key from. Also, if a configured key file is not
18520 found the tool will now automatically fall back to prompting
18523 * Python systemd.journal module was updated to wrap recently
18524 added functions from libsystemd-journal. The interface was
18525 changed to bring the low level interface in s.j._Reader
18526 closer to the C API, and the high level interface in
18527 s.j.Reader was updated to wrap and convert all data about
18530 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
18531 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
18532 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
18533 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
18534 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
18535 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
18539 * journalctl --update-catalog now understands a new --root=
18540 option to operate on catalogs found in a different root
18543 * During shutdown after systemd has terminated all running
18544 services a final killing loop kills all remaining left-over
18545 processes. We will now print the name of these processes
18546 when we send SIGKILL to them, since this usually indicates a
18549 * If /etc/crypttab refers to password files stored on
18550 configured mount points automatic dependencies will now be
18551 generated to ensure the specific mount is established first
18552 before the key file is attempted to be read.
18554 * 'systemctl status' will now show information about the
18555 network sockets a socket unit is listening on.
18557 * 'systemctl status' will also shown information about any
18558 drop-in configuration file for units. (Drop-In configuration
18559 files in this context are files such as
18560 /etc/systemd/system/foobar.service.d/*.conf)
18562 * systemd-cgtop now optionally shows summed up CPU times of
18563 cgroups. Press '%' while running cgtop to switch between
18564 percentage and absolute mode. This is useful to determine
18565 which cgroups use up the most CPU time over the entire
18566 runtime of the system. systemd-cgtop has also been updated
18567 to be 'pipeable' for processing with further shell tools.
18569 * 'hostnamectl set-hostname' will now allow setting of FQDN
18572 * The formatting and parsing of time span values has been
18573 changed. The parser now understands fractional expressions
18574 such as "5.5h". The formatter will now output fractional
18575 expressions for all time spans under 1min, i.e. "5.123456s"
18576 rather than "5s 123ms 456us". For time spans under 1s
18577 millisecond values are shown, for those under 1ms
18578 microsecond values are shown. This should greatly improve
18579 all time-related output of systemd.
18581 * libsystemd-login and libsystemd-journal gained new
18582 functions for querying the poll() events mask and poll()
18583 timeout value for integration into arbitrary event
18586 * localectl gained the ability to list available X11 keymaps
18587 (models, layouts, variants, options).
18589 * 'systemd-analyze dot' gained the ability to filter for
18590 specific units via shell-style globs, to create smaller,
18591 more useful graphs. I.e. it is now possible to create simple
18592 graphs of all the dependencies between only target units, or
18593 of all units that Avahi has dependencies with.
18595 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
18596 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
18597 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
18598 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
18599 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
18600 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
18601 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
18605 * The boot-time readahead implementation for rotating media
18606 will now read the read-ahead data in multiple passes which
18607 consist of all read requests made in equidistant time
18608 intervals. This means instead of strictly reading read-ahead
18609 data in its physical order on disk we now try to find a
18610 middle ground between physical and access time order.
18612 * /etc/os-release files gained a new BUILD_ID= field for usage
18613 on operating systems that provide continuous builds of OS
18616 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
18617 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
18618 William Douglas, Zbigniew Jędrzejewski-Szmek
18622 * systemd-python gained an API exposing libsystemd-daemon.
18624 * The SMACK setup logic gained support for uploading CIPSO
18627 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
18628 ReadOnlyDirectories= and InaccessibleDirectories= has
18629 changed. The private /tmp and /var/tmp directories are now
18630 shared by all processes of a service (which means
18631 ExecStartPre= may now leave data in /tmp that ExecStart= of
18632 the same service can still access). When a service is
18633 stopped its temporary directories are immediately deleted
18634 (normal clean-up with tmpfiles is still done in addition to
18637 * By default, systemd will now set a couple of sysctl
18638 variables in the kernel: the safe sysrq options are turned
18639 on, IP route verification is turned on, and source routing
18640 disabled. The recently added hardlink and softlink
18641 protection of the kernel is turned on. These settings should
18642 be reasonably safe, and good defaults for all new systems.
18644 * The predictable network naming logic may now be turned off
18645 with a new kernel command line switch: net.ifnames=0.
18647 * A new libsystemd-bus module has been added that implements a
18648 pretty complete D-Bus client library. For details see:
18650 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
18652 * journald will now explicitly flush the journal files to disk
18653 at the latest 5min after each write. The file will then also
18654 be marked offline until the next write. This should increase
18655 reliability in case of a crash. The synchronization delay
18656 can be configured via SyncIntervalSec= in journald.conf.
18658 * There's a new remote-fs-setup.target unit that can be used
18659 to pull in specific services when at least one remote file
18660 system is to be mounted.
18662 * There are new targets timers.target and paths.target as
18663 canonical targets to pull user timer and path units in
18664 from. This complements sockets.target with a similar
18665 purpose for socket units.
18667 * libudev gained a new call udev_device_set_attribute_value()
18668 to set sysfs attributes of a device.
18670 * The udev daemon now sets the default number of worker
18671 processes executed in parallel based on the number of available
18672 CPUs instead of the amount of available RAM. This is supposed
18673 to provide a more reliable default and limit a too aggressive
18674 parallelism for setups with 1000s of devices connected.
18676 Contributions from: Auke Kok, Colin Walters, Cristian
18677 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
18678 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
18679 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
18680 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
18681 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
18682 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
18683 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
18684 Zbigniew Jędrzejewski-Szmek
18688 * Configuration of unit files may now be extended via drop-in
18689 files without having to edit/override the unit files
18690 themselves. More specifically, if the administrator wants to
18691 change one value for a service file foobar.service he can
18692 now do so by dropping in a configuration snippet into
18693 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
18694 will load all these snippets and apply them on top of the
18695 main unit configuration file, possibly extending or
18696 overriding its settings. Using these drop-in snippets is
18697 generally nicer than the two earlier options for changing
18698 unit files locally: copying the files from
18699 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
18700 them there; or creating a new file in /etc/systemd/system/
18701 that incorporates the original one via ".include". Drop-in
18702 snippets into these .d/ directories can be placed in any
18703 directory systemd looks for units in, and the usual
18704 overriding semantics between /usr/lib, /etc and /run apply
18707 * Most unit file settings which take lists of items can now be
18708 reset by assigning the empty string to them. For example,
18709 normally, settings such as Environment=FOO=BAR append a new
18710 environment variable assignment to the environment block,
18711 each time they are used. By assigning Environment= the empty
18712 string the environment block can be reset to empty. This is
18713 particularly useful with the .d/*.conf drop-in snippets
18714 mentioned above, since this adds the ability to reset list
18715 settings from vendor unit files via these drop-ins.
18717 * systemctl gained a new "list-dependencies" command for
18718 listing the dependencies of a unit recursively.
18720 * Inhibitors are now honored and listed by "systemctl
18721 suspend", "systemctl poweroff" (and similar) too, not only
18722 GNOME. These commands will also list active sessions by
18725 * Resource limits (as exposed by the various control group
18726 controllers) can now be controlled dynamically at runtime
18727 for all units. More specifically, you can now use a command
18728 like "systemctl set-cgroup-attr foobar.service cpu.shares
18729 2000" to alter the CPU shares a specific service gets. These
18730 settings are stored persistently on disk, and thus allow the
18731 administrator to easily adjust the resource usage of
18732 services with a few simple commands. This dynamic resource
18733 management logic is also available to other programs via the
18734 bus. Almost any kernel cgroup attribute and controller is
18737 * systemd-vconsole-setup will now copy all font settings to
18738 all allocated VTs, where it previously applied them only to
18741 * libsystemd-login gained the new sd_session_get_tty() API
18744 * This release drops support for a few legacy or
18745 distribution-specific LSB facility names when parsing init
18746 scripts: $x-display-manager, $mail-transfer-agent,
18747 $mail-transport-agent, $mail-transfer-agent, $smtp,
18748 $null. Also, the mail-transfer-agent.target unit backing
18749 this has been removed. Distributions which want to retain
18750 compatibility with this should carry the burden for
18751 supporting this themselves and patch support for these back
18752 in, if they really need to. Also, the facilities $syslog and
18753 $local_fs are now ignored, since systemd does not support
18754 early-boot LSB init scripts anymore, and these facilities
18755 are implied anyway for normal services. syslog.target has
18758 * There are new bus calls on PID1's Manager object for
18759 cancelling jobs, and removing snapshot units. Previously,
18760 both calls were only available on the Job and Snapshot
18761 objects themselves.
18763 * systemd-journal-gatewayd gained SSL support.
18765 * The various "environment" files, such as /etc/locale.conf
18766 now support continuation lines with a backslash ("\") as
18767 last character in the line, similarly in style (but different)
18768 to how this is supported in shells.
18770 * For normal user processes the _SYSTEMD_USER_UNIT= field is
18771 now implicitly appended to every log entry logged. systemctl
18772 has been updated to filter by this field when operating on a
18773 user systemd instance.
18775 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
18776 CAP_AUDIT_CONTROL capabilities to the capabilities set for
18777 the container. This makes it easier to boot unmodified
18778 Fedora systems in a container, which however still requires
18779 audit=0 to be passed on the kernel command line. Auditing in
18780 kernel and userspace is unfortunately still too broken in
18781 context of containers, hence we recommend compiling it out
18782 of the kernel or using audit=0. Hopefully this will be fixed
18783 one day for good in the kernel.
18785 * nspawn gained the new --bind= and --bind-ro= parameters to
18786 bind mount specific directories from the host into the
18789 * nspawn will now mount its own devpts file system instance
18790 into the container, in order not to leak pty devices from
18791 the host into the container.
18793 * systemd will now read the firmware boot time performance
18794 information from the EFI variables, if the used boot loader
18795 supports this, and takes it into account for boot performance
18796 analysis via "systemd-analyze". This is currently supported
18797 only in conjunction with Gummiboot, but could be supported
18798 by other boot loaders too. For details see:
18800 https://systemd.io/BOOT_LOADER_INTERFACE
18802 * A new generator has been added that automatically mounts the
18803 EFI System Partition (ESP) to /boot, if that directory
18804 exists, is empty, and no other file system has been
18805 configured to be mounted there.
18807 * logind will now send out PrepareForSleep(false) out
18808 unconditionally, after coming back from suspend. This may be
18809 used by applications as asynchronous notification for
18810 system resume events.
18812 * "systemctl unlock-sessions" has been added, that allows
18813 unlocking the screens of all user sessions at once, similar
18814 to how "systemctl lock-sessions" already locked all users
18815 sessions. This is backed by a new D-Bus call UnlockSessions().
18817 * "loginctl seat-status" will now show the master device of a
18818 seat. (i.e. the device of a seat that needs to be around for
18819 the seat to be considered available, usually the graphics
18822 * tmpfiles gained a new "X" line type, that allows
18823 configuration of files and directories (with wildcards) that
18824 shall be excluded from automatic cleanup ("aging").
18826 * udev default rules set the device node permissions now only
18827 at "add" events, and do not change them any longer with a
18828 later "change" event.
18830 * The log messages for lid events and power/sleep keypresses
18831 now carry a message ID.
18833 * We now have a substantially larger unit test suite, but this
18834 continues to be work in progress.
18836 * udevadm hwdb gained a new --root= parameter to change the
18837 root directory to operate relative to.
18839 * logind will now issue a background sync() request to the kernel
18840 early at shutdown, so that dirty buffers are flushed to disk early
18841 instead of at the last moment, in order to optimize shutdown
18844 * A new bootctl tool has been added that is an interface for
18845 certain boot loader operations. This is currently a preview
18846 and is likely to be extended into a small mechanism daemon
18847 like timedated, localed, hostnamed, and can be used by
18848 graphical UIs to enumerate available boot options, and
18849 request boot into firmware operations.
18851 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
18852 the rest of the package. It also has been updated to work
18853 correctly in initrds.
18855 * polkit previously has been runtime optional, and is now also
18856 compile time optional via a configure switch.
18858 * systemd-analyze has been reimplemented in C. Also "systemctl
18859 dot" has moved into systemd-analyze.
18861 * "systemctl status" with no further parameters will now print
18862 the status of all active or failed units.
18864 * Operations such as "systemctl start" can now be executed
18865 with a new mode "--irreversible" which may be used to queue
18866 operations that cannot accidentally be reversed by a later
18867 job queuing. This is by default used to make shutdown
18868 requests more robust.
18870 * The Python API of systemd now gained a new module for
18871 reading journal files.
18873 * A new tool kernel-install has been added that can install
18874 kernel images according to the Boot Loader Specification:
18876 https://systemd.io/BOOT_LOADER_SPECIFICATION
18878 * Boot time console output has been improved to provide
18879 animated boot time output for hanging jobs.
18881 * A new tool systemd-activate has been added which can be used
18882 to test socket activation with, directly from the command
18883 line. This should make it much easier to test and debug
18884 socket activation in daemons.
18886 * journalctl gained a new "--reverse" (or -r) option to show
18887 journal output in reverse order (i.e. newest line first).
18889 * journalctl gained a new "--pager-end" (or -e) option to jump
18890 to immediately jump to the end of the journal in the
18891 pager. This is only supported in conjunction with "less".
18893 * journalctl gained a new "--user-unit=" option, that works
18894 similarly to "--unit=" but filters for user units rather than
18897 * A number of unit files to ease adoption of systemd in
18898 initrds has been added. This moves some minimal logic from
18899 the various initrd implementations into systemd proper.
18901 * The journal files are now owned by a new group
18902 "systemd-journal", which exists specifically to allow access
18903 to the journal, and nothing else. Previously, we used the
18904 "adm" group for that, which however possibly covers more
18905 than just journal/log file access. This new group is now
18906 already used by systemd-journal-gatewayd to ensure this
18907 daemon gets access to the journal files and as little else
18908 as possible. Note that "make install" will also set FS ACLs
18909 up for /var/log/journal to give "adm" and "wheel" read
18910 access to it, in addition to "systemd-journal" which owns
18911 the journal files. We recommend that packaging scripts also
18912 add read access to "adm" + "wheel" to /var/log/journal, and
18913 all existing/future journal files. To normal users and
18914 administrators little changes, however packagers need to
18915 ensure to create the "systemd-journal" system group at
18916 package installation time.
18918 * The systemd-journal-gatewayd now runs as unprivileged user
18919 systemd-journal-gateway:systemd-journal-gateway. Packaging
18920 scripts need to create these system user/group at
18923 * timedated now exposes a new boolean property CanNTP that
18924 indicates whether a local NTP service is available or not.
18926 * systemd-detect-virt will now also detect xen PVs
18928 * The pstore file system is now mounted by default, if it is
18931 * In addition to the SELinux and IMA policies we will now also
18932 load SMACK policies at early boot.
18934 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
18935 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
18936 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
18937 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
18938 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
18939 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
18940 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
18941 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
18942 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
18943 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
18944 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
18945 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
18946 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
18947 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
18951 * Timer units now support calendar time events in addition to
18952 monotonic time events. That means you can now trigger a unit
18953 based on a calendar time specification such as "Thu,Fri
18954 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
18955 or fifth day of any month of the year 2013, given that it is
18956 a Thursday or a Friday. This brings timer event support
18957 considerably closer to cron's capabilities. For details on
18958 the supported calendar time specification language see
18961 * udev now supports a number of different naming policies for
18962 network interfaces for predictable names, and a combination
18963 of these policies is now the default. Please see this wiki
18964 document for details:
18966 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
18968 * Auke Kok's bootchart implementation has been added to the
18969 systemd tree. It is an optional component that can graph the
18970 boot in quite some detail. It is one of the best bootchart
18971 implementations around and minimal in its code and
18974 * nss-myhostname has been integrated into the systemd source
18975 tree. nss-myhostname guarantees that the local hostname
18976 always stays resolvable via NSS. It has been a weak
18977 requirement of systemd-hostnamed since a long time, and
18978 since its code is actually trivial we decided to just
18979 include it in systemd's source tree. It can be turned off
18980 with a configure switch.
18982 * The read-ahead logic is now capable of properly detecting
18983 whether a btrfs file system is on SSD or rotating media, in
18984 order to optimize the read-ahead scheme. Previously, it was
18985 only capable of detecting this on traditional file systems
18988 * In udev, additional device properties are now read from the
18989 IAB in addition to the OUI database. Also, Bluetooth company
18990 identities are attached to the devices as well.
18992 * In service files %U may be used as specifier that is
18993 replaced by the configured user name of the service.
18995 * nspawn may now be invoked without a controlling TTY. This
18996 makes it suitable for invocation as its own service. This
18997 may be used to set up a simple containerized server system
18998 using only core OS tools.
19000 * systemd and nspawn can now accept socket file descriptors
19001 when they are started for socket activation. This enables
19002 implementation of socket activated nspawn
19003 containers. i.e. think about autospawning an entire OS image
19004 when the first SSH or HTTP connection is received. We expect
19005 that similar functionality will also be added to libvirt-lxc
19008 * journalctl will now suppress ANSI color codes when
19009 presenting log data.
19011 * systemctl will no longer show control group information for
19012 a unit if the control group is empty anyway.
19014 * logind can now automatically suspend/hibernate/shutdown the
19017 * /etc/machine-info and hostnamed now also expose the chassis
19018 type of the system. This can be used to determine whether
19019 the local system is a laptop, desktop, handset or
19020 tablet. This information may either be configured by the
19021 user/vendor or is automatically determined from ACPI and DMI
19022 information if possible.
19024 * A number of polkit actions are now bound together with "imply"
19025 rules. This should simplify creating UIs because many actions
19026 will now authenticate similar ones as well.
19028 * Unit files learnt a new condition ConditionACPower= which
19029 may be used to conditionalize a unit depending on whether an
19030 AC power source is connected or not, of whether the system
19031 is running on battery power.
19033 * systemctl gained a new "is-failed" verb that may be used in
19034 shell scripts and suchlike to check whether a specific unit
19035 is in the "failed" state.
19037 * The EnvironmentFile= setting in unit files now supports file
19038 globbing, and can hence be used to easily read a number of
19039 environment files at once.
19041 * systemd will no longer detect and recognize specific
19042 distributions. All distribution-specific #ifdeffery has been
19043 removed, systemd is now fully generic and
19044 distribution-agnostic. Effectively, not too much is lost as
19045 a lot of the code is still accessible via explicit configure
19046 switches. However, support for some distribution specific
19047 legacy configuration file formats has been dropped. We
19048 recommend distributions to simply adopt the configuration
19049 files everybody else uses now and convert the old
19050 configuration from packaging scripts. Most distributions
19051 already did that. If that's not possible or desirable,
19052 distributions are welcome to forward port the specific
19053 pieces of code locally from the git history.
19055 * When logging a message about a unit systemd will now always
19056 log the unit name in the message meta data.
19058 * localectl will now also discover system locale data that is
19059 not stored in locale archives, but directly unpacked.
19061 * logind will no longer unconditionally use framebuffer
19062 devices as seat masters, i.e. as devices that are required
19063 to be existing before a seat is considered preset. Instead,
19064 it will now look for all devices that are tagged as
19065 "seat-master" in udev. By default, framebuffer devices will
19066 be marked as such, but depending on local systems, other
19067 devices might be marked as well. This may be used to
19068 integrate graphics cards using closed source drivers (such
19069 as NVidia ones) more nicely into logind. Note however, that
19070 we recommend using the open source NVidia drivers instead,
19071 and no udev rules for the closed-source drivers will be
19072 shipped from us upstream.
19074 Contributions from: Adam Williamson, Alessandro Crismani, Auke
19075 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
19076 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
19077 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
19078 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
19079 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
19080 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
19081 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
19082 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
19083 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
19084 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
19085 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
19090 * udev gained support for loading additional device properties
19091 from an indexed database that is keyed by vendor/product IDs
19092 and similar device identifiers. For the beginning this
19093 "hwdb" is populated with data from the well-known PCI and
19094 USB database, but also includes PNP, ACPI and OID data. In
19095 the longer run this indexed database shall grow into
19096 becoming the one central database for non-essential
19097 userspace device metadata. Previously, data from the PCI/USB
19098 database was only attached to select devices, since the
19099 lookup was a relatively expensive operation due to O(n) time
19100 complexity (with n being the number of entries in the
19101 database). Since this is now O(1), we decided to add in this
19102 data for all devices where this is available, by
19103 default. Note that the indexed database needs to be rebuilt
19104 when new data files are installed. To achieve this you need
19105 to update your packaging scripts to invoke "udevadm hwdb
19106 --update" after installation of hwdb data files. For
19107 RPM-based distributions we introduced the new
19108 %udev_hwdb_update macro for this purpose.
19110 * The Journal gained support for the "Message Catalog", an
19111 indexed database to link up additional information with
19112 journal entries. For further details please check:
19114 https://systemd.io/CATALOG
19116 The indexed message catalog database also needs to be
19117 rebuilt after installation of message catalog files. Use
19118 "journalctl --update-catalog" for this. For RPM-based
19119 distributions we introduced the %journal_catalog_update
19120 macro for this purpose.
19122 * The Python Journal bindings gained support for the standard
19123 Python logging framework.
19125 * The Journal API gained new functions for checking whether
19126 the underlying file system of a journal file is capable of
19127 properly reporting file change notifications, or whether
19128 applications that want to reflect journal changes "live"
19129 need to recheck journal files continuously in appropriate
19132 * It is now possible to set the "age" field for tmpfiles
19133 entries to 0, indicating that files matching this entry
19134 shall always be removed when the directories are cleaned up.
19136 * coredumpctl gained a new "gdb" verb which invokes gdb
19137 right-away on the selected coredump.
19139 * There's now support for "hybrid sleep" on kernels that
19140 support this, in addition to "suspend" and "hibernate". Use
19141 "systemctl hybrid-sleep" to make use of this.
19143 * logind's HandleSuspendKey= setting (and related settings)
19144 now gained support for a new "lock" setting to simply
19145 request the screen lock on all local sessions, instead of
19146 actually executing a suspend or hibernation.
19148 * systemd will now mount the EFI variables file system by
19151 * Socket units now gained support for configuration of the
19152 SMACK security label.
19154 * timedatectl will now output the time of the last and next
19155 daylight saving change.
19157 * We dropped support for various legacy and distro-specific
19158 concepts, such as insserv, early-boot SysV services
19159 (i.e. those for non-standard runlevels such as 'b' or 'S')
19160 or ArchLinux /etc/rc.conf support. We recommend the
19161 distributions who still need support this to either continue
19162 to maintain the necessary patches downstream, or find a
19163 different solution. (Talk to us if you have questions!)
19165 * Various systemd components will now bypass polkit checks for
19166 root and otherwise handle properly if polkit is not found to
19167 be around. This should fix most issues for polkit-less
19168 systems. Quite frankly this should have been this way since
19169 day one. It is absolutely our intention to make systemd work
19170 fine on polkit-less systems, and we consider it a bug if
19171 something does not work as it should if polkit is not around.
19173 * For embedded systems it is now possible to build udev and
19174 systemd without blkid and/or kmod support.
19176 * "systemctl switch-root" is now capable of switching root
19177 more than once. I.e. in addition to transitions from the
19178 initrd to the host OS it is now possible to transition to
19179 further OS images from the host. This is useful to implement
19180 offline updating tools.
19182 * Various other additions have been made to the RPM macros
19183 shipped with systemd. Use %udev_rules_update() after
19184 installing new udev rules files. %_udevhwdbdir,
19185 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
19186 %_sysctldir are now available which resolve to the right
19187 directories for packages to place various data files in.
19189 * journalctl gained the new --full switch (in addition to
19190 --all, to disable ellipsation for long messages.
19192 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
19193 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
19194 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
19195 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
19196 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
19197 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
19198 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
19199 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
19200 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
19204 * journalctl gained new --since= and --until= switches to
19205 filter by time. It also now supports nice filtering for
19206 units via --unit=/-u.
19208 * Type=oneshot services may use ExecReload= and do the
19211 * The journal daemon now supports time-based rotation and
19212 vacuuming, in addition to the usual disk-space based
19215 * The journal will now index the available field values for
19216 each field name. This enables clients to show pretty drop
19217 downs of available match values when filtering. The bash
19218 completion of journalctl has been updated
19219 accordingly. journalctl gained a new switch -F to list all
19220 values a certain field takes in the journal database.
19222 * More service events are now written as structured messages
19223 to the journal, and made recognizable via message IDs.
19225 * The timedated, localed and hostnamed mini-services which
19226 previously only provided support for changing time, locale
19227 and hostname settings from graphical DEs such as GNOME now
19228 also have a minimal (but very useful) text-based client
19229 utility each. This is probably the nicest way to changing
19230 these settings from the command line now, especially since
19231 it lists available options and is fully integrated with bash
19234 * There's now a new tool "systemd-coredumpctl" to list and
19235 extract coredumps from the journal.
19237 * We now install a README each in /var/log/ and
19238 /etc/rc.d/init.d explaining where the system logs and init
19239 scripts went. This hopefully should help folks who go to
19240 that dirs and look into the otherwise now empty void and
19241 scratch their heads.
19243 * When user-services are invoked (by systemd --user) the
19244 $MANAGERPID env var is set to the PID of systemd.
19246 * SIGRTMIN+24 when sent to a --user instance will now result
19247 in immediate termination of systemd.
19249 * gatewayd received numerous feature additions such as a
19250 "follow" mode, for live syncing and filtering.
19252 * browse.html now allows filtering and showing detailed
19253 information on specific entries. Keyboard navigation and
19254 mouse screen support has been added.
19256 * gatewayd/journalctl now supports HTML5/JSON
19257 Server-Sent-Events as output.
19259 * The SysV init script compatibility logic will now
19260 heuristically determine whether a script supports the
19261 "reload" verb, and only then make this available as
19262 "systemctl reload".
19264 * "systemctl status --follow" has been removed, use "journalctl
19267 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
19268 have been removed since they are hardly useful to be
19271 * And I'd like to take the opportunity to specifically mention
19272 Zbigniew for his great contributions. Zbigniew, you rock!
19274 Contributions from: Andrew Eikum, Christian Hesse, Colin
19275 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
19276 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
19277 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
19278 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
19279 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
19280 Jędrzejewski-Szmek, Сковорода Никита Андреевич
19284 * If /etc/vconsole.conf is non-existent or empty we will no
19285 longer load any console font or key map at boot by
19286 default. Instead the kernel defaults will be left
19287 intact. This is definitely the right thing to do, as no
19288 configuration should mean no configuration, and hard-coding
19289 font names that are different on all archs is probably a bad
19290 idea. Also, the kernel default key map and font should be
19291 good enough for most cases anyway, and mostly identical to
19292 the userspace fonts/key maps we previously overloaded them
19293 with. If distributions want to continue to default to a
19294 non-kernel font or key map they should ship a default
19295 /etc/vconsole.conf with the appropriate contents.
19297 Contributions from: Colin Walters, Daniel J Walsh, Dave
19298 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
19299 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
19303 * journalctl gained a new --cursor= switch to show entries
19304 starting from the specified location in the journal.
19306 * We now enforce a size limit on journal entry fields exported
19307 with "-o json" in journalctl. Fields larger than 4K will be
19308 assigned null. This can be turned off with --all.
19310 * An (optional) journal gateway daemon is now available as
19311 "systemd-journal-gatewayd.service". This service provides
19312 access to the journal via HTTP and JSON. This functionality
19313 will be used to implement live log synchronization in both
19314 pull and push modes, but has various other users too, such
19315 as easy log access for debugging of embedded devices. Right
19316 now it is already useful to retrieve the journal via HTTP:
19318 # systemctl start systemd-journal-gatewayd.service
19319 # wget http://localhost:19531/entries
19321 This will download the journal contents in a
19322 /var/log/messages compatible format. The same as JSON:
19324 # curl -H"Accept: application/json" http://localhost:19531/entries
19326 This service is also accessible via a web browser where a
19327 single static HTML5 app is served that uses the JSON logic
19328 to enable the user to do some basic browsing of the
19329 journal. This will be extended later on. Here's an example
19330 screenshot of this app in its current state:
19332 https://0pointer.de/public/journal-gatewayd
19334 Contributions from: Kay Sievers, Lennart Poettering, Robert
19335 Milasan, Tom Gundersen
19339 * The bash completion logic is now available for journalctl
19342 * We do not mount the "cpuset" controller anymore together with
19343 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
19344 started if no parameters are assigned to it. "cpuset" hence
19345 broke code that assumed it could create "cpu" groups and
19348 * journalctl -f will now subscribe to terminal size changes,
19349 and line break accordingly.
19351 Contributions from: Dave Reisner, Kay Sievers, Lennart
19352 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
19356 * nspawn will now create a symlink /etc/localtime in the
19357 container environment, copying the host's timezone
19358 setting. Previously this has been done via a bind mount, but
19359 since symlinks cannot be bind mounted this has now been
19360 changed to create/update the appropriate symlink.
19362 * journalctl -n's line number argument is now optional, and
19363 will default to 10 if omitted.
19365 * journald will now log the maximum size the journal files may
19366 take up on disk. This is particularly useful if the default
19367 built-in logic of determining this parameter from the file
19368 system size is used. Use "systemctl status
19369 systemd-journald.service" to see this information.
19371 * The multi-seat X wrapper tool has been stripped down. As X
19372 is now capable of enumerating graphics devices via udev in a
19373 seat-aware way the wrapper is not strictly necessary
19374 anymore. A stripped down temporary stop-gap is still shipped
19375 until the upstream display managers have been updated to
19376 fully support the new X logic. Expect this wrapper to be
19377 removed entirely in one of the next releases.
19379 * HandleSleepKey= in logind.conf has been split up into
19380 HandleSuspendKey= and HandleHibernateKey=. The old setting
19381 is not available anymore. X11 and the kernel are
19382 distinguishing between these keys and we should too. This
19383 also means the inhibition lock for these keys has been split
19386 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
19387 Poettering, Lukas Nykryn, Václav Pavlín
19391 * Whenever a unit changes state we will now log this to the
19392 journal and show along the unit's own log output in
19393 "systemctl status".
19395 * ConditionPathIsMountPoint= can now properly detect bind
19396 mount points too. (Previously, a bind mount of one file
19397 system to another place in the same file system could not be
19398 detected as mount, since they shared struct stat's st_dev
19401 * We will now mount the cgroup controllers cpu, cpuacct,
19402 cpuset and the controllers net_cls, net_prio together by
19405 * nspawn containers will now have a virtualized boot
19406 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
19407 over with a randomized ID at container initialization). This
19408 has the effect of making "journalctl -b" do the right thing
19411 * The JSON output journal serialization has been updated not
19412 to generate "endless" list objects anymore, but rather one
19413 JSON object per line. This is more in line how most JSON
19414 parsers expect JSON objects. The new output mode
19415 "json-pretty" has been added to provide similar output, but
19416 neatly aligned for readability by humans.
19418 * We dropped all explicit sync() invocations in the shutdown
19419 code. The kernel does this implicitly anyway in the kernel
19420 reboot() syscall. halt(8)'s -n option is now a compatibility
19423 * We now support virtualized reboot() in containers, as
19424 supported by newer kernels. We will fall back to exit() if
19425 CAP_SYS_REBOOT is not available to the container. Also,
19426 nspawn makes use of this now and will actually reboot the
19427 container if the containerized OS asks for that.
19429 * journalctl will only show local log output by default
19430 now. Use --merge (-m) to show remote log output, too.
19432 * libsystemd-journal gained the new sd_journal_get_usage()
19433 call to determine the current disk usage of all journal
19434 files. This is exposed in the new "journalctl --disk-usage"
19437 * journald gained a new configuration setting SplitMode= in
19438 journald.conf which may be used to control how user journals
19439 are split off. See journald.conf(5) for details.
19441 * A new condition type ConditionFileNotEmpty= has been added.
19443 * tmpfiles' "w" lines now support file globbing, to write
19444 multiple files at once.
19446 * We added Python bindings for the journal submission
19447 APIs. More Python APIs for a number of selected APIs will
19448 likely follow. Note that we intend to add native bindings
19449 only for the Python language, as we consider it common
19450 enough to deserve bindings shipped within systemd. There are
19451 various projects outside of systemd that provide bindings
19452 for languages such as PHP or Lua.
19454 * Many conditions will now resolve specifiers such as %i. In
19455 addition, PathChanged= and related directives of .path units
19456 now support specifiers as well.
19458 * There's now a new RPM macro definition for the system preset
19461 * journald will now warn if it ca not forward a message to the
19462 syslog daemon because its socket is full.
19464 * timedated will no longer write or process /etc/timezone,
19465 except on Debian. As we do not support late mounted /usr
19466 anymore /etc/localtime always being a symlink is now safe,
19467 and hence the information in /etc/timezone is not necessary
19470 * logind will now always reserve one VT for a text getty (VT6
19471 by default). Previously if more than 6 X sessions where
19472 started they took up all the VTs with auto-spawned gettys,
19473 so that no text gettys were available anymore.
19475 * udev will now automatically inform the btrfs kernel logic
19476 about btrfs RAID components showing up. This should make
19477 simple hotplug based btrfs RAID assembly work.
19479 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
19480 (but not for its children which will stay at the kernel
19481 default). This should allow setups with a lot more listening
19484 * systemd will now always pass the configured timezone to the
19485 kernel at boot. timedated will do the same when the timezone
19488 * logind's inhibition logic has been updated. By default,
19489 logind will now handle the lid switch, the power and sleep
19490 keys all the time, even in graphical sessions. If DEs want
19491 to handle these events on their own they should take the new
19492 handle-power-key, handle-sleep-key and handle-lid-switch
19493 inhibitors during their runtime. A simple way to achieve
19494 that is to invoke the DE wrapped in an invocation of:
19496 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
19498 * Access to unit operations is now checked via SELinux taking
19499 the unit file label and client process label into account.
19501 * systemd will now notify the administrator in the journal
19502 when he over-mounts a non-empty directory.
19504 * There are new specifiers that are resolved in unit files,
19505 for the hostname (%H), the machine ID (%m) and the boot ID
19508 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
19509 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
19510 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
19511 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
19512 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
19513 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
19514 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
19518 * Support for reading structured kernel messages from
19519 /dev/kmsg has now been added and is enabled by default.
19521 * Support for reading kernel messages from /proc/kmsg has now
19522 been removed. If you want kernel messages in the journal
19523 make sure to run a recent kernel (>= 3.5) that supports
19524 reading structured messages from /dev/kmsg (see
19525 above). /proc/kmsg is now exclusive property of classic
19526 syslog daemons again.
19528 * The libudev API gained the new
19529 udev_device_new_from_device_id() call.
19531 * The logic for file system namespace (ReadOnlyDirectory=,
19532 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
19533 require pivot_root() anymore. This means fewer temporary
19534 directories are created below /tmp for this feature.
19536 * nspawn containers will now see and receive all submounts
19537 made on the host OS below the root file system of the
19540 * Forward Secure Sealing is now supported for Journal files,
19541 which provide cryptographical sealing of journal files so
19542 that attackers cannot alter log history anymore without this
19543 being detectable. Lennart will soon post a blog story about
19544 this explaining it in more detail.
19546 * There are two new service settings RestartPreventExitStatus=
19547 and SuccessExitStatus= which allow configuration of exit
19548 status (exit code or signal) which will be excepted from the
19549 restart logic, resp. consider successful.
19551 * journalctl gained the new --verify switch that can be used
19552 to check the integrity of the structure of journal files and
19553 (if Forward Secure Sealing is enabled) the contents of
19556 * nspawn containers will now be run with /dev/stdin, /dev/fd/
19557 and similar symlinks pre-created. This makes running shells
19558 as container init process a lot more fun.
19560 * The fstab support can now handle PARTUUID= and PARTLABEL=
19563 * A new ConditionHost= condition has been added to match
19564 against the hostname (with globs) and machine ID. This is
19565 useful for clusters where a single OS image is used to
19566 provision a large number of hosts which shall run slightly
19567 different sets of services.
19569 * Services which hit the restart limit will now be placed in a
19572 Contributions from: Bertram Poettering, Dave Reisner, Huang
19573 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
19574 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
19578 * When running in --user mode systemd will now become a
19579 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
19580 tree a lot more organized.
19582 * A new PartOf= unit dependency type has been introduced that
19583 may be used to group services in a natural way.
19585 * "systemctl enable" may now be used to enable instances of
19588 * journalctl now prints error log levels in red, and
19589 warning/notice log levels in bright white. It also supports
19590 filtering by log level now.
19592 * cgtop gained a new -n switch (similar to top), to configure
19593 the maximum number of iterations to run for. It also gained
19594 -b, to run in batch mode (accepting no input).
19596 * The suffix ".service" may now be omitted on most systemctl
19597 command lines involving service unit names.
19599 * There's a new bus call in logind to lock all sessions, as
19600 well as a loginctl verb for it "lock-sessions".
19602 * libsystemd-logind.so gained a new call sd_journal_perror()
19603 that works similar to libc perror() but logs to the journal
19604 and encodes structured information about the error number.
19606 * /etc/crypttab entries now understand the new keyfile-size=
19609 * shutdown(8) now can send a (configurable) wall message when
19610 a shutdown is cancelled.
19612 * The mount propagation mode for the root file system will now
19613 default to "shared", which is useful to make containers work
19614 nicely out-of-the-box so that they receive new mounts from
19615 the host. This can be undone locally by running "mount
19616 --make-rprivate /" if needed.
19618 * The prefdm.service file has been removed. Distributions
19619 should maintain this unit downstream if they intend to keep
19620 it around. However, we recommend writing normal unit files
19621 for display managers instead.
19623 * Since systemd is a crucial part of the OS we will now
19624 default to a number of compiler switches that improve
19625 security (hardening) such as read-only relocations, stack
19626 protection, and suchlike.
19628 * The TimeoutSec= setting for services is now split into
19629 TimeoutStartSec= and TimeoutStopSec= to allow configuration
19630 of individual time outs for the start and the stop phase of
19633 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
19634 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
19635 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
19636 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
19637 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
19638 Gundersen, Zbigniew Jędrzejewski-Szmek
19642 * The journal and id128 C APIs are now fully documented as man
19645 * Extra safety checks have been added when transitioning from
19646 the initial RAM disk to the main system to avoid accidental
19649 * /etc/crypttab entries now understand the new keyfile-offset=
19652 * systemctl -t can now be used to filter by unit load state.
19654 * The journal C API gained the new sd_journal_wait() call to
19655 make writing synchronous journal clients easier.
19657 * journalctl gained the new -D switch to show journals from a
19658 specific directory.
19660 * journalctl now displays a special marker between log
19661 messages of two different boots.
19663 * The journal is now explicitly flushed to /var via a service
19664 systemd-journal-flush.service, rather than implicitly simply
19665 by seeing /var/log/journal to be writable.
19667 * journalctl (and the journal C APIs) can now match for much
19668 more complex expressions, with alternatives and
19671 * When transitioning from the initial RAM disk to the main
19672 system we will now kill all processes in a killing spree to
19673 ensure no processes stay around by accident.
19675 * Three new specifiers may be used in unit files: %u, %h, %s
19676 resolve to the user name, user home directory resp. user
19677 shell. This is useful for running systemd user instances.
19679 * We now automatically rotate journal files if their data
19680 object hash table gets a fill level > 75%. We also size the
19681 hash table based on the configured maximum file size. This
19682 together should lower hash collisions drastically and thus
19683 speed things up a bit.
19685 * journalctl gained the new "--header" switch to introspect
19686 header data of journal files.
19688 * A new setting SystemCallFilters= has been added to services which may
19689 be used to apply deny lists or allow lists to system calls. This is
19690 based on SECCOMP Mode 2 of Linux 3.5.
19692 * nspawn gained a new --link-journal= switch (and quicker: -j)
19693 to link the container journal with the host. This makes it
19694 very easy to centralize log viewing on the host for all
19695 guests while still keeping the journal files separated.
19697 * Many bugfixes and optimizations
19699 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
19700 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
19701 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
19706 * Several tools now understand kernel command line arguments,
19707 which are only read when run in an initial RAM disk. They
19708 usually follow closely their normal counterparts, but are
19711 * There's a new tool to analyze the readahead files that are
19712 automatically generated at boot. Use:
19714 /usr/lib/systemd/systemd-readahead analyze /.readahead
19716 * We now provide an early debug shell on tty9 if this enabled. Use:
19718 systemctl enable debug-shell.service
19720 * All plymouth related units have been moved into the Plymouth
19721 package. Please make sure to upgrade your Plymouth version
19724 * systemd-tmpfiles now supports getting passed the basename of
19725 a configuration file only, in which case it will look for it
19726 in all appropriate directories automatically.
19728 * udevadm info now takes a /dev or /sys path as argument, and
19729 does the right thing. Example:
19731 udevadm info /dev/sda
19732 udevadm info /sys/class/block/sda
19734 * systemctl now prints a warning if a unit is stopped but a
19735 unit that might trigger it continues to run. Example: a
19736 service is stopped but the socket that activates it is left
19739 * "systemctl status" will now mention if the log output was
19740 shortened due to rotation since a service has been started.
19742 * The journal API now exposes functions to determine the
19743 "cutoff" times due to rotation.
19745 * journald now understands SIGUSR1 and SIGUSR2 for triggering
19746 immediately flushing of runtime logs to /var if possible,
19747 resp. for triggering immediate rotation of the journal
19750 * It is now considered an error if a service is attempted to
19751 be stopped that is not loaded.
19753 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
19755 * systemd-analyze now supports Python 3
19757 * tmpfiles now supports cleaning up directories via aging
19758 where the first level dirs are always kept around but
19759 directories beneath it automatically aged. This is enabled
19760 by prefixing the age field with '~'.
19762 * Seat objects now expose CanGraphical, CanTTY properties
19763 which is required to deal with very fast bootups where the
19764 display manager might be running before the graphics drivers
19765 completed initialization.
19767 * Seat objects now expose a State property.
19769 * We now include RPM macros for service enabling/disabling
19770 based on the preset logic. We recommend RPM based
19771 distributions to make use of these macros if possible. This
19772 makes it simpler to reuse RPM spec files across
19775 * We now make sure that the collected systemd unit name is
19776 always valid when services log to the journal via
19779 * There's a new man page kernel-command-line(7) detailing all
19780 command line options we understand.
19782 * The fstab generator may now be disabled at boot by passing
19783 fstab=0 on the kernel command line.
19785 * A new kernel command line option modules-load= is now understood
19786 to load a specific kernel module statically, early at boot.
19788 * Unit names specified on the systemctl command line are now
19789 automatically escaped as needed. Also, if file system or
19790 device paths are specified they are automatically turned
19791 into the appropriate mount or device unit names. Example:
19793 systemctl status /home
19794 systemctl status /dev/sda
19796 * The SysVConsole= configuration option has been removed from
19797 system.conf parsing.
19799 * The SysV search path is no longer exported on the D-Bus
19802 * The Names= option has been removed from unit file parsing.
19804 * There's a new man page bootup(7) detailing the boot process.
19806 * Every unit and every generator we ship with systemd now
19807 comes with full documentation. The self-explanatory boot is
19810 * A couple of services gained "systemd-" prefixes in their
19811 name if they wrap systemd code, rather than only external
19812 code. Among them fsck@.service which is now
19813 systemd-fsck@.service.
19815 * The HaveWatchdog property has been removed from the D-Bus
19818 * systemd.confirm_spawn= on the kernel command line should now
19821 * There's a new man page crypttab(5) which details all options
19822 we actually understand.
19824 * systemd-nspawn gained a new --capability= switch to pass
19825 additional capabilities to the container.
19827 * timedated will now read known NTP implementation unit names
19828 from /usr/lib/systemd/ntp-units.d/*.list,
19829 systemd-timedated-ntp.target has been removed.
19831 * journalctl gained a new switch "-b" that lists log data of
19832 the current boot only.
19834 * The notify socket is in the abstract namespace again, in
19835 order to support daemons which chroot() at start-up.
19837 * There is a new Storage= configuration option for journald
19838 which allows configuration of where log data should go. This
19839 also provides a way to disable journal logging entirely, so
19840 that data collected is only forwarded to the console, the
19841 kernel log buffer or another syslog implementation.
19843 * Many bugfixes and optimizations
19845 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
19846 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
19847 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
19848 Shawn Landden, Tom Gundersen
19852 * "systemctl help <unit>" now shows the man page if one is
19855 * Several new man pages have been added.
19857 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
19858 MaxLevelConsole= can now be specified in
19859 journald.conf. These options allow reducing the amount of
19860 data stored on disk or forwarded by the log level.
19862 * TimerSlackNSec= can now be specified in system.conf for
19863 PID1. This allows system-wide power savings.
19865 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
19866 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
19871 * logind is now capable of (optionally) handling power and
19872 sleep keys as well as the lid switch.
19874 * journalctl now understands the syntax "journalctl
19875 /usr/bin/avahi-daemon" to get all log output of a specific
19878 * CapabilityBoundingSet= in system.conf now also influences
19879 the capability bound set of usermode helpers of the kernel.
19881 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
19882 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
19883 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
19884 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
19888 * Note that we skipped 139 releases here in order to set the
19889 new version to something that is greater than both udev's
19890 and systemd's most recent version number.
19892 * udev: all udev sources are merged into the systemd source tree now.
19893 All future udev development will happen in the systemd tree. It
19894 is still fully supported to use the udev daemon and tools without
19895 systemd running, like in initramfs or other init systems. Building
19896 udev though, will require the *build* of the systemd tree, but
19897 udev can be properly *run* without systemd.
19899 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
19900 should be used to create dead device nodes as workarounds for broken
19903 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
19904 no longer supported. udev_monitor_new_from_netlink() needs to be
19905 used to subscribe to events.
19907 * udev: when udevd is started by systemd, processes which are left
19908 behind by forking them off of udev rules, are unconditionally cleaned
19909 up and killed now after the event handling has finished. Services or
19910 daemons must be started as systemd services. Services can be
19911 pulled-in by udev to get started, but they can no longer be directly
19912 forked by udev rules.
19914 * udev: the daemon binary is called systemd-udevd now and installed
19915 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
19916 to adapt to that, create symlink, or rename the binary after building
19919 * libudev no longer provides these symbols:
19920 udev_monitor_from_socket()
19921 udev_queue_get_failed_list_entry()
19922 udev_get_{dev,sys,run}_path()
19923 The versions number was bumped and symbol versioning introduced.
19925 * systemd-loginctl and systemd-journalctl have been renamed
19926 to loginctl and journalctl to match systemctl.
19928 * The config files: /etc/systemd/systemd-logind.conf and
19929 /etc/systemd/systemd-journald.conf have been renamed to
19930 logind.conf and journald.conf. Package updates should rename
19931 the files to the new names on upgrade.
19933 * For almost all files the license is now LGPL2.1+, changed
19934 from the previous GPL2.0+. Exceptions are some minor stuff
19935 of udev (which will be changed to LGPL2.1 eventually, too),
19936 and the MIT licensed sd-daemon.[ch] library that is suitable
19937 to be used as drop-in files.
19939 * systemd and logind now handle system sleep states, in
19940 particular suspending and hibernating.
19942 * logind now implements a sleep/shutdown/idle inhibiting logic
19943 suitable for a variety of uses. Soonishly Lennart will blog
19944 about this in more detail.
19946 * var-run.mount and var-lock.mount are no longer provided
19947 (which previously bind mounted these directories to their new
19948 places). Distributions which have not converted these
19949 directories to symlinks should consider stealing these files
19950 from git history and add them downstream.
19952 * We introduced the Documentation= field for units and added
19953 this to all our shipped units. This is useful to make it
19954 easier to explore the boot and the purpose of the various
19957 * All smaller setup units (such as
19958 systemd-vconsole-setup.service) now detect properly if they
19959 are run in a container and are skipped when
19960 appropriate. This guarantees an entirely noise-free boot in
19961 Linux container environments such as systemd-nspawn.
19963 * A framework for implementing offline system updates is now
19964 integrated, for details see:
19965 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
19967 * A new service type Type=idle is available now which helps us
19968 avoiding ugly interleaving of getty output and boot status
19971 * There's now a system-wide CapabilityBoundingSet= option to
19972 globally reduce the set of capabilities for the
19973 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
19974 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
19975 even CAP_NET_ADMIN system-wide for secure systems.
19977 * There are now system-wide DefaultLimitXXX= options to
19978 globally change the defaults of the various resource limits
19979 for all units started by PID 1.
19981 * Harald Hoyer's systemd test suite has been integrated into
19982 systemd which allows easy testing of systemd builds in qemu
19983 and nspawn. (This is really awesome! Ask us for details!)
19985 * The fstab parser is now implemented as generator, not inside
19988 * systemctl will now warn you if .mount units generated from
19989 /etc/fstab are out of date due to changes in fstab that
19990 have not been read by systemd yet.
19992 * systemd is now suitable for usage in initrds. Dracut has
19993 already been updated to make use of this. With this in place
19994 initrds get a slight bit faster but primarily are much
19995 easier to introspect and debug since "systemctl status" in
19996 the host system can be used to introspect initrd services,
19997 and the journal from the initrd is kept around too.
19999 * systemd-delta has been added, a tool to explore differences
20000 between user/admin configuration and vendor defaults.
20002 * PrivateTmp= now affects both /tmp and /var/tmp.
20004 * Boot time status messages are now much prettier and feature
20005 proper english language. Booting up systemd has never been
20008 * Read-ahead pack files now include the inode number of all
20009 files to pre-cache. When the inode changes the pre-caching
20010 is not attempted. This should be nicer to deal with updated
20011 packages which might result in changes of read-ahead
20014 * We now temporaritly lower the kernel's read_ahead_kb variable
20015 when collecting read-ahead data to ensure the kernel's
20016 built-in read-ahead does not add noise to our measurements
20017 of necessary blocks to pre-cache.
20019 * There's now RequiresMountsFor= to add automatic dependencies
20020 for all mounts necessary for a specific file system path.
20022 * MountAuto= and SwapAuto= have been removed from
20023 system.conf. Mounting file systems at boot has to take place
20026 * nspawn now learned a new switch --uuid= to set the machine
20027 ID on the command line.
20029 * nspawn now learned the -b switch to automatically search
20030 for an init system.
20032 * vt102 is now the default TERM for serial TTYs, upgraded from
20035 * systemd-logind now works on VT-less systems.
20037 * The build tree has been reorganized. The individual
20038 components now have directories of their own.
20040 * A new condition type ConditionPathIsReadWrite= is now available.
20042 * nspawn learned the new -C switch to create cgroups for the
20043 container in other hierarchies.
20045 * We now have support for hardware watchdogs, configurable in
20048 * The scheduled shutdown logic now has a public API.
20050 * We now mount /tmp as tmpfs by default, but this can be
20051 masked and /etc/fstab can override it.
20053 * Since udisks does not make use of /media anymore we are not
20054 mounting a tmpfs on it anymore.
20056 * journalctl gained a new --local switch to only interleave
20057 locally generated journal files.
20059 * We can now load the IMA policy at boot automatically.
20061 * The GTK tools have been split off into a systemd-ui.
20063 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
20064 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
20065 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
20066 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
20067 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
20068 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
20069 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
20070 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
20071 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
20076 * This is mostly a bugfix release
20078 * Support optional initialization of the machine ID from the
20079 KVM or container configured UUID.
20081 * Support immediate reboots with "systemctl reboot -ff"
20083 * Show /etc/os-release data in systemd-analyze output
20085 * Many bugfixes for the journal, including endianness fixes and
20086 ensuring that disk space enforcement works
20088 * sd-login.h is C++ compatible again
20090 * Extend the /etc/os-release format on request of the Debian
20093 * We now refuse non-UTF8 strings used in various configuration
20094 and unit files. This is done to ensure we do not pass invalid
20095 data over D-Bus or expose it elsewhere.
20097 * Register Mimo USB Screens as suitable for automatic seat
20100 * Read SELinux client context from journal clients in a race
20103 * Reorder configuration file lookup order. /etc now always
20104 overrides /run in order to allow the administrator to always
20105 and unconditionally override vendor-supplied or
20106 automatically generated data.
20108 * The various user visible bits of the journal now have man
20109 pages. We still lack man pages for the journal API calls
20112 * We now ship all man pages in HTML format again in the
20115 Contributions from: Dave Reisner, Dirk Eibach, Frederic
20116 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
20117 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
20122 * This is mostly a bugfix release
20124 * systems lacking /etc/os-release are no longer supported.
20126 * Various functionality updates to libsystemd-login.so
20128 * Track class of PAM logins to distinguish greeters from
20129 normal user logins.
20131 Contributions from: Kay Sievers, Lennart Poettering, Michael
20136 * This is an important bugfix release for v41.
20138 * Building man pages is now optional which should be useful
20139 for those building systemd from git but unwilling to install
20142 * Watchdog support for supervising services is now usable. In
20143 a future release support for hardware watchdogs
20144 (i.e. /dev/watchdog) will be added building on this.
20146 * Service start rate limiting is now configurable and can be
20147 turned off per service. When a start rate limit is hit a
20148 reboot can automatically be triggered.
20150 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
20152 Contributions from: Benjamin Franzke, Bill Nottingham,
20153 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
20154 Schmidt, Michał Górny, Piotr Drąg
20158 * The systemd binary is installed /usr/lib/systemd/systemd now;
20159 An existing /sbin/init symlink needs to be adapted with the
20162 * The code that loads kernel modules has been ported to invoke
20163 libkmod directly, instead of modprobe. This means we do not
20164 support systems with module-init-tools anymore.
20166 * Watchdog support is now already useful, but still not
20169 * A new kernel command line option systemd.setenv= is
20170 understood to set system wide environment variables
20171 dynamically at boot.
20173 * We now limit the set of capabilities of systemd-journald.
20175 * We now set SIGPIPE to ignore by default, since it only is
20176 useful in shell pipelines, and has little use in general
20177 code. This can be disabled with IgnoreSIPIPE=no in unit
20180 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
20181 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
20186 * This is mostly a bugfix release
20188 * We now expose the reason why a service failed in the
20189 "Result" D-Bus property.
20191 * Rudimentary service watchdog support (will be completed over
20192 the next few releases.)
20194 * When systemd forks off in order execute some service we will
20195 now immediately changes its argv[0] to reflect which process
20196 it will execute. This is useful to minimize the time window
20197 with a generic argv[0], which makes bootcharts more useful
20199 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
20200 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
20201 Mike Kazantsev, Ray Strode
20205 * This is mostly a test release, but incorporates many
20208 * New systemd-cgtop tool to show control groups by their
20211 * Linking against libacl for ACLs is optional again. If
20212 disabled, support tracking device access for active logins
20213 goes becomes unavailable, and so does access to the user
20214 journals by the respective users.
20216 * If a group "adm" exists, journal files are automatically
20217 owned by them, thus allow members of this group full access
20218 to the system journal as well as all user journals.
20220 * The journal now stores the SELinux context of the logging
20221 client for all entries.
20223 * Add C++ inclusion guards to all public headers
20225 * New output mode "cat" in the journal to print only text
20226 messages, without any meta data like date or time.
20228 * Include tiny X server wrapper as a temporary stop-gap to
20229 teach XOrg udev display enumeration. This is used by display
20230 managers such as gdm, and will go away as soon as XOrg
20231 learned native udev hotplugging for display devices.
20233 * Add new systemd-cat tool for executing arbitrary programs
20234 with STDERR/STDOUT connected to the journal. Can also act as
20235 BSD logger replacement, and does so by default.
20237 * Optionally store all locally generated coredumps in the
20238 journal along with meta data.
20240 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
20241 writing short strings to files (for usage for /sys), and for
20242 creating symlinks, character and block device nodes.
20244 * New unit file option ControlGroupPersistent= to make cgroups
20245 persistent, following the mechanisms outlined in
20246 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
20248 * Support multiple local RTCs in a sane way
20250 * No longer monopolize IO when replaying readahead data on
20251 rotating disks, since we might starve non-file-system IO to
20252 death, since fanotify() will not see accesses done by blkid,
20255 * Do not show kernel threads in systemd-cgls anymore, unless
20256 requested with new -k switch.
20258 Contributions from: Dan Horák, Kay Sievers, Lennart
20259 Poettering, Michal Schmidt
20263 * This is mostly a test release, but incorporates many
20266 * The git repository moved to:
20267 git://anongit.freedesktop.org/systemd/systemd
20268 ssh://git.freedesktop.org/git/systemd/systemd
20270 * First release with the journal
20271 https://0pointer.de/blog/projects/the-journal.html
20273 * The journal replaces both systemd-kmsg-syslogd and
20274 systemd-stdout-bridge.
20276 * New sd_pid_get_unit() API call in libsystemd-logind
20278 * Many systemadm clean-ups
20280 * Introduce remote-fs-pre.target which is ordered before all
20281 remote mounts and may be used to start services before all
20284 * Added Mageia support
20286 * Add bash completion for systemd-loginctl
20288 * Actively monitor PID file creation for daemons which exit in
20289 the parent process before having finished writing the PID
20290 file in the daemon process. Daemons which do this need to be
20291 fixed (i.e. PID file creation must have finished before the
20292 parent exits), but we now react a bit more gracefully to them.
20294 * Add colourful boot output, mimicking the well-known output
20295 of existing distributions.
20297 * New option PassCredentials= for socket units, for
20298 compatibility with a recent kernel ABI breakage.
20300 * /etc/rc.local is now hooked in via a generator binary, and
20301 thus will no longer act as synchronization point during
20304 * systemctl list-unit-files now supports --root=.
20306 * systemd-tmpfiles now understands two new commands: z, Z for
20307 relabelling files according to the SELinux database. This is
20308 useful to apply SELinux labels to specific files in /sys,
20309 among other things.
20311 * Output of SysV services is now forwarded to both the console
20312 and the journal by default, not only just the console.
20314 * New man pages for all APIs from libsystemd-login.
20316 * The build tree got reorganized and the build system is a
20317 lot more modular allowing embedded setups to specifically
20318 select the components of systemd they are interested in.
20320 * Support for Linux systems lacking the kernel VT subsystem is
20323 * configure's --with-rootdir= got renamed to
20324 --with-rootprefix= to follow the naming used by udev and
20327 * Unless specified otherwise we will now install to /usr instead
20328 of /usr/local by default.
20330 * Processes with '@' in argv[0][0] are now excluded from the
20331 final shut-down killing spree, following the logic explained
20333 https://systemd.io/ROOT_STORAGE_DAEMONS/
20335 * All processes remaining in a service cgroup when we enter
20336 the START or START_PRE states are now killed with
20337 SIGKILL. That means it is no longer possible to spawn
20338 background processes from ExecStart= lines (which was never
20339 supported anyway, and bad style).
20341 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
20342 reloading of units together.
20344 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
20345 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
20346 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
20347 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
20348 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / thirdparty / systemd.git / blob