Environment=ONE='one' "TWO='two two' too" THREE=
ExecStart=/bin/python3 -c 'import sys;print(sys.argv)' $ONE $TWO $THREE
-* When systemctl --host is used, underlying ssh connection can remain open.
- bus_close does not kill children?
-
External:
* Fedora: add an rpmlint check that verifies that all unit files in the RPM are listed in %systemd_post macros.
Janitorial Clean-ups:
-* replace manual readdir() loops with FOREACH_DIRENT or FOREACH_DIRENT_ALL
-
* Rearrange tests so that the various test-xyz.c match a specific src/basic/xyz.c again
Features:
+* add some special mode to LogsDirectory=/StateDirectory=… that allows
+ declaring these directories without necessarily pulling in deps for them, or
+ creating them when starting up. That way, we could declare that
+ systemd-journald writes to /var/log/journal, which could be useful when we
+ doing disk usage calculations and so on.
+
+* taint systemd if the overflowuid/overflowgid is not 65534
+
+* deprecate PermissionsStartOnly= and RootDirectoryStartOnly= in favour of the ExecStart= prefix chars
+
+* add a new RuntimeDirectoryPreserve= mode that defines a similar lifecycle for
+ the runtime dir as we maintain for the fdstore: i.e. keep it around as long
+ as the unit is running or has a job queued.
+
+* hook up sd-bus' creds stuff with SO_PEERGROUPS
+
+* add async version of sd_bus_add_match and make use of that
+
+* support projid-based quota in machinectl for containers, and then drop
+ implicit btrfs loopback magic in machined
+
+* let's log the "tainted" string at boot
+
+* Add NetworkNamespacePath= to specify a path to a network namespace
+
+* maybe use SOURCE_DATE_EPOCH (i.e. the env var the reproducible builds folks
+ introduced) as the RTC epoch, instead of the mtime of NEWS.
+
+* add a way to lock down cgroup migration: a boolean, which when set for a unit
+ makes sure the processes in it can never migrate out of it
+
+* blog about fd store and restartable services
+
+* document Environment=SYSTEMD_LOG_LEVEL=debug drop-in in debugging document
+
+* rework ExecOutput and ExecInput enums so that EXEC_OUTPUT_NULL loses its
+ magic meaning and is no longer upgraded to something else if set explicitly.
+
+* in the long run: permit a system with /etc/machine-id linked to /dev/null, to
+ make it lose its identity, i.e. be anonymous. For this we'd have to patch
+ through the whole tree to make all code deal with the case where no machine
+ ID is available.
+
+* optionally, collect cgroup resource data, and store it in per-unit RRD files,
+ suitable for processing with rrdtool. Add bus API to access this data, and
+ possibly implement a CPULoad property based on it.
+
+* beef up pam_systemd to take unit file settings such as cgroups properties as
+ parameters
+
+* export UID ranges nspawns's --private-user and DynamicUser= uses in
+ the systemd.pc pkg-config file, the same way we already expose the system
+ user boundary there
+
+* a new "systemd-analyze security" tool outputting a checklist of security
+ features a service does and does not implement
+
+* Whenever we check a UID against the system UID range, also check for the
+ dynamic UID range
+
+* maybe hook of xfs/ext4 quotactl() with services? i.e. automatically manage
+ the quota of a the user indicated in User= via unit file settings, like the
+ other resource management concepts. Would mix nicely with DynamicUser=1. Or
+ alternatively, do this with projids, so that we can also cover services
+ running as root. Quota should probably cover all the special dirs such as
+ StateDirectory=, LogsDirectory=, CacheDirectory=, as well as RootDirectory= if it
+ is set, plus the whole disk space any image configured with RootImage=.
+
+* Introduce "exit" as an EmergencyAction value, and allow to configure a
+ per-unit success/failure exit code to configure. This would be useful for
+ running commands inside of services inside of containers, which could then
+ propagate their failure state all the way up.
+
+* In DynamicUser= mode: before selecting a UID, use disk quota APIs on relevant
+ disks to see if the UID is already in use.
+
+* add dissect_image_warn() as a wrapper around dissect_image() that prints
+ friendly log messages for the returned errors, so that we don't have to
+ duplicate that in nspawn, systemd-dissect and PID 1.
+
+* add "systemctl wait" or so, which does what "systemd-run --wait" does, but
+ for all units. It should be both a way to pin units into memory as well as a
+ wait to retrieve their exit data.
+
+* maybe set a new set of env vars for services, based on RuntimeDirectory=,
+ StateDirectory=, LogsDirectory=, CacheDirectory= and ConfigurationDirectory=
+ automatically. For example, there could be $RUNTIME_DIRECTORY,
+ $STATE_DIRECTORY, $LOGS_DIRECTORY=, $CACHE_DIRECTORY and
+ $CONFIGURATION_DIRECTORY or so. This could be useful to write services that
+ can adapt to varying directories for these purposes. Special care has to be
+ taken if multiple dirs are configured. Maybe avoid setting the env vars in
+ that case?
+
+* expose IO accounting data on the bus, show it in systemd-run --wait and log
+ about it in the resource log message
+
+* rework unbase64 code to drop whitespace automatically, so that we don't have
+ to drop it first.
+
+* add "systemctl purge" for flushing out configuration, state, logs, ... of a
+ unit when it is stopped
+
+* show whether a service has out-of-date configuration in "systemctl status" by
+ using mtime data of ConfigurationDirectory=.
+
+* replace all uses of fgets() + LINE_MAX by read_line()
+
+* Add AddUser= setting to unit files, similar to DynamicUser=1 which however
+ creates a static, persistent user rather than a dynamic, transient user. We
+ can leverage code from sysusers.d for this.
+
+* add some optional flag to ReadWritePaths= and friends, that has the effect
+ that we create the dir in question when the service is started. Example:
+
+ ReadWritePaths=:/var/lib/foobar
+
+* maybe add call sd_journal_set_block_timeout() or so to set SO_SNDTIMEO for
+ the sd-journal logging socket, and, if the timeout is set to 0, sets
+ O_NONBLOCK on it. That way people can control if and when to block for
+ logging.
+
+* tighten sd_notify() MAINPID= checks a bit: don't accept foreign PIDs (i.e.
+ PIDs not managed by the service manager)
+
+* hostnamed: populate form factor data from a new hwdb database, so that old
+ yogas can be recognized as "convertible" too, even if they predate the DMI
+ "convertible" form factor
+
+* Maybe add a small tool invoked early at boot, that adds in or resizes
+ partitions automatically, to be used when the media used is actually larger
+ than the image written onto it is.
+
+* Maybe add PrivatePIDs= as new unit setting, and do minimal PID namespacing
+ after all. Be strict however, only support the equivalent of nspawn's
+ --as-pid2 switch, and sanely proxy sd_notify() messages dropping stuff such
+ as MAINPID.
+
+* Add ExecMonitor= setting. May be used multiple times. Forks off a process in
+ the service cgroup, which is supposed to monitor the service, and when it
+ exits the service is considered failed by its monitor.
+
+* track the per-service PAM process properly (i.e. as an additional control
+ process), so that it may be queried on the bus and everything.
+
+* add a new "debug" job mode, that is propagated to unit_start() and for
+ services results in two things: we raise SIGSTOP right before invoking
+ execve() and turn off watchdog support. Then, use that to implement
+ "systemd-gdb" for attaching to the start-up of any system service in its
+ natural habitat.
+
* replace all canonicalize_file_name() invocations by chase_symlinks(), in
particulr those where a rootdir is relevant.
-* drop nss-myhostname in favour of nss-resolve?
+* maybe introduce gpt auto discovery for /var/tmp?
+
+* maybe add gpt-partition-based user management: each user gets his own
+ LUKS-encrypted GPT partition with a new GPT type. A small nss module
+ enumerates users via udev partition enumeration. UIDs are assigned in a fixed
+ way: the partition index is added as offset to some fixed base uid. User name
+ is stored in GPT partition name. A PAM module authenticates the user via the
+ LUKS partition password. Benefits: strong per-user security, compatibility
+ with stateless/read-only/verity-enabled root. (other idea: do this based on
+ loopback files in /home, without GPT involvement)
-* drop internal dlopen() based nss-dns fallback in nss-resolve, and rely on the
- external nsswitch.conf based one
+* gpt-auto logic: introduce support for discovering /var matching an image. For
+ that, use a partition type UUID that is hashed from the OS name (as encoded
+ in /etc/os-release), the architecture, and 4 new bits from the gpt flags
+ field of the root partition. This way can easily support multiple OS
+ installations on the same GPT partition table, without problems with
+ unmatched /var partitions.
+
+* gpt-auto logic: related to the above, maybe support a "secondary" root
+ partition, that is mounted to / and is writable, and where the actual root's
+ /usr is mounted into.
+
+* .mount and .swap units: add Format=yes|no option that formats the partition before mounting/enabling it, implicitly
+
+* gpt-auto logic: support encrypted swap, add kernel cmdline option to force it, and honour a gpt bit about it, plus maybe a configuration file
+
+* drop nss-myhostname in favour of nss-resolve?
* add a percentage syntax for TimeoutStopSec=, e.g. TimeoutStopSec=150%, and
then use that for the setting used in user@.service. It should be understood
relative to the configured default value.
-* on cgroupsv2 add DelegateControllers=, to pick the precise cgroup controllers to delegate
-
* in networkd, when matching device types, fix up DEVTYPE rubbish the kernel passes to us
* enable LockMLOCK to take a percentage value relative to physical memory
-* switch to ProtectSystem=strict for all our long-running services where that's possible
-
-* If RootDirectory= is used, mount /proc, /sys, /dev into it, if not mounted yet
-
* Permit masking specific netlink APIs with RestrictAddressFamily=
-* nspawn: start UID allocation loop from hash of container name
-
* nspawn: support that /proc, /sys/, /dev are pre-mounted
* define gpt header bits to select volatility mode
* DeviceAllow= should also generate seccomp filters for mknod()
-* Add DataDirectory=, CacheDirectory= and LogDirectory= to match
- RuntimeDirectory=, and create it as necessary when starting a service, owned by the right user.
-
-* Add BindDirectory= for allowing arbitrary, private bind mounts for services
-
-* Add RootImage= for mounting a disk image or file as root directory
-
* make sure the ratelimit object can deal with USEC_INFINITY as way to turn off things
* journalctl: make sure -f ends when the container indicated by -M terminates
a user/group for a service only has to exist on the host for the right
mapping to work.
-* allow attaching additional journald log fields to cgroups
-
* add bus API for creating unit files in /etc, reusing the code for transient units
* add bus API to remove unit files from /etc
the specified range and generates sane error messages for incorrect
specifications.
-* do something about "/control" subcgroups in the unified cgroup hierarchy
-
* when we detect that there are waiting jobs but no running jobs, do something
* push CPUAffinity= also into the "cpuset" cgroup controller (only after the cpuset controller got ported to the unified hierarchy)
prefixed with /sys generally special.
http://lists.freedesktop.org/archives/systemd-devel/2015-June/032962.html
-* man: document that unless you use StandardError=null the shell >/dev/stderr won't work in shell scripts in services
-
* fstab-generator: default to tmpfs-as-root if only usr= is specified on the kernel cmdline
* docs: bring http://www.freedesktop.org/wiki/Software/systemd/MyServiceCantGetRealtime up to date
* Rework systemctl's GetAll property parsing to use the generic bus_map_all_properties() API
-* implement a per-service firewall based on net_cls
-
* Port various tools to make use of verbs.[ch], where applicable: busctl,
coredumpctl, hostnamectl, localectl, systemd-analyze, timedatectl
* introduce systemd-timesync-wait.service or so to sync on an NTP fix?
-* systemd --user should issue sd_notify() upon reaching basic.target, not on becoming idle
-
* consider showing the unit names during boot up in the status output, not just the unit descriptions
* maybe allow timer units with an empty Units= setting, so that they
* support empty /etc boots nicely:
- nspawn/gpt-generator: introduce new gpt partition type for /usr
- - fstab-generator: support systemd.volatile=yes|no|state on the kernel cmdline, too, similar to nspawn's --volatile=
* generator that automatically discovers btrfs subvolumes, identifies their purpose based on some xattr on them.
* For timer units: add some mechanisms so that timer units that trigger immediately on boot do not have the services
they run added to the initial transaction and thus confuse Type=idle.
-* Run most system services with cgroupfs read-only and procfs with a more secure mode (doesn't work, since the hidepid= option is per-pid-namespace, not per-mount)
-
* add bus api to query unit file's X fields.
* gpt-auto-generator:
- - Support LUKS for root devices
- Define new partition type for encrypted swap? Support probed LUKS for encrypted swap?
- Make /home automount rather than mount?
* refuse boot if /usr/lib/os-release is missing or /etc/machine-id cannot be set up
-* btrfs raid assembly: some .device jobs stay stuck in the queue
-
* man: the documentation of Restart= currently is very misleading and suggests the tools from ExecStartPre= might get restarted.
* load .d/*.conf dropins for device units
* sd-bus:
- EBADSLT handling
- GetAllProperties() on a non-existing object does not result in a failure currently
- - kdbus: process fd=-1 for incoming msgs
- port to sd-resolve for connecting to TCP dbus servers
- - kdbus: maybe add controlling tty metadata fields
- see if we can introduce a new sd_bus_get_owner_machine_id() call to retrieve the machine ID of the machine of the bus itself
- - when kdbus does not take our message without memfds, try again with memfds
- see if we can drop more message validation on the sending side
- add API to clone sd_bus_message objects
- make AddMatch calls on dbus1 transports async?
- - kdbus: matches against source or destination pids for an "strace -p"-like feel. Problem: The PID info needs to be available in userspace too...
- longer term: priority inheritance
- dbus spec updates:
- - kdbus mapping
- NameLost/NameAcquired obsolete
- GVariant
- path escaping
- update systemd.special(7) to mention that dbus.socket is only about the compatibility socket now
- test bloom filter generation indexes
- - kdbus: introduce a concept of "send-only" connections
- - kdbus: add counter for refused unicast messages that is passed out via the RECV ioctl. SImilar to the counter for dropped multicast messages we already have.
* sd-event
- allow multiple signal handlers per signal?
* maybe add a generator that looks for "systemd.run=" on the kernel cmdline for container usercases...
-* cgtop: make cgtop useful in a container
-
* test/:
- add 'set -e' to scripts in test/
- make stuff in test/ work with separate output dir
* shutdown logging: store to EFI var, and store to USB stick?
-* think about window-manager-run-as-user-service problem: exit 0 → activate shutdown.target; exit != 0 → restart service
-
* merge unit_kill_common() and unit_kill_context()
* introduce ExecCondition= in services
- journald: when we drop syslog messages because the syslog socket is
full, make sure to write how many messages are lost as first thing
to syslog when it works again.
- - journald: make sure ratelimit is actually really per-service with the new cgroup changes
- change systemd-journal-flush into a service that stays around during
boot, and causes the journal to be moved back to /run on shutdown,
so that we do not keep /var busy. This needs to happen synchronously,
- add journalctl -H that talks via ssh to a remote peer and passes through
binary logs data
- add a version of --merge which also merges /var/log/journal/remote
- - log accumulated resource usage after each service invocation
- journalctl: -m should access container journals directly by enumerating
them via machined, and also watch containers coming and going.
Benefit: nspawn --ephemeral would start working nicely with the journal.
- document that deps in [Unit] sections ignore Alias= fields in
[Install] units of other units, unless those units are disabled
- man: clarify that time-sync.target is not only sysv compat but also useful otherwise. Same for similar targets
- - document the exit codes when services fail before they are exec()ed
- document that service reload may be implemented as service reexec
- - document in wiki how to map ical recurrence events to systemd timer unit calendar specifications
- add a man page containing packaging guidelines and recommending usage of things like Documentation=, PrivateTmp=, PrivateNetwork= and ReadOnlyDirectories=/etc /usr.
- document systemd-journal-flush.service properly
- documentation: recommend to connect the timer units of a service to the service via Also= in [Install]
- add new command to systemctl: "systemctl system-reexec" which reexecs as many daemons as virtually possible
- systemctl enable: fail if target to alias into does not exist? maybe show how many units are enabled afterwards?
- systemctl: "Journal has been rotated since unit was started." message is misleading
- - better error message if you run systemctl without systemd running
- systemctl status output should include list of triggering units and their status
* unit install:
* on shutdown: move utmp, wall, audit logic all into PID 1 (or logind?), get rid of systemd-update-utmp-runlevel
-* make repeated alt-ctrl-del presses printing a dump, or even force a reboot without
- waiting for the timeout
+* make repeated alt-ctrl-del presses printing a dump
* hostnamed: before returning information from /etc/machine-info.conf check the modification data and reread. Similar for localed, ...
* currently x-systemd.timeout is lost in the initrd, since crypttab is copied into dracut, but fstab is not
* nspawn:
- - nspawn -x should support ephemeral instances of gpt images
- emulate /dev/kmsg using CUSE and turn off the syslog syscall
with seccomp. That should provide us with a useful log buffer that
systemd can log to during early boot, and disconnect container logs
- should send out sd_notify("WATCHDOG=1") messages
- optionally automatically add FORWARD rules to iptables whenever nspawn is
running, remove them when shut down.
- - Improve error message when --bind= is used on a non-existing source
- directory
- maybe make copying of /etc/resolv.conf optional, and skip it if --read-only
is used
+* dissect
+ - refuse mounting over a mount point
+ - automatically discover .roothash files in dissect, similarly to nspawn
+
* machined:
- add an API so that libvirt-lxc can inform us about network interfaces being
removed or added to an existing machine
* cryptsetup:
- cryptsetup-generator: allow specification of passwords in crypttab itself
- - move cryptsetup key caching into kernel keyctl?
- https://bugs.freedesktop.org/show_bug.cgi?id=54982
- support rd.luks.allow-discards= kernel cmdline params in cryptsetup generator
* hw watchdog: optionally try to use the preset watchdog timeout instead of always overriding it
* create /sbin/init symlinks from the build system
+* add a dependency on standard-conf.xml and other included files to man pages
+
* MountFlags=shared acts as MountFlags=slave right now.
* properly handle loop back mounts via fstab, especially regards to fsck/passno
- add trigger --subsystem-match=usb/usb_device device
- reimport udev db after MOVE events for devices without dev_t
-* when a service has the same env var set twice we actually store it twice and return that in systemctl show -p... We should only show the last setting
-
* There's currently no way to cancel fsck (used to be possible via C-c or c on the console)
* add option to sockets to avoid activation. Instead just drop packets/connections, see http://cyberelk.net/tim/2012/02/15/portreserve-systemd-solution/
* write blog stories about:
- hwdb: what belongs into it, lsusb
- enabling dbus services
- - status update
- how to make changes to sysctl and sysfs attributes
- remote access
- how to pass throw-away units to systemd, or dynamically change properties of existing units
* dot output for --test showing the 'initial transaction'
-* fingerprint.target, wireless.target, gps.target, netdevice.target
-
* pid1:
- - .timer units should optionally support CLOCK_BOOTTIME in addition to CLOCK_MONOTONIC
- When logging about multiple units (stopping BoundTo units, conflicts, etc.),
log both units as UNIT=, so that journalctl -u triggers on both.
- generate better errors when people try to set transient properties
- load-fragment: when loading a unit file via a chain of symlinks
verify that it is not masked via any of the names traversed.
- introduce Type=pid-file
- - ExecOnFailure=/usr/bin/foo
- introduce mix of BindTo and Requisite
- add a concept of RemainAfterExit= to scope units
- - Set NoNewPrivileges= on all of our own services, where that makes sense
- Allow multiple ExecStart= for all Type= settings, so that we can cover rescue.service nicely
- - consider adding RuntimeDirectoryUser= + RuntimeDirectoryGroup=
* udev-link-config:
- Make sure ID_PATH is always exported and complete for
* dhcp:
- figure out how much we can increase Maximum Message Size
- - support RFC4702 (pass FQDN)
* dhcp6:
- add functions to set previously stored IPv6 addresses on startup and get
* drop accountsservice's StandardOutput=syslog and Type=dbus fields
-* dbus: in fedora, make /var/lib/dbus/machine-id a symlink to /etc/machine-id
-
* /usr/bin/service should actually show the new command line
* fedora: suggest auto-restart on failure, but not on success and not on coredump. also, ask people to think about changing the start limit logic. Also point people to RestartPreventExitStatus=, SuccessExitStatus=
* check for strerror(r) instead of strerror(-r)
-* Use PR_SET_PROCTITLE_AREA if it becomes available in the kernel
-
* pahole
* set_put(), hashmap_put() return values check. i.e. == 0 does not free()!