3 * Many manager configuration settings that are only applicable to user
4 manager or system manager can be always set. It would be better to reject
5 them when parsing config.
7 * Jun 01 09:43:02 krowka systemd[1]: Unit user@1000.service has alias user@.service.
8 Jun 01 09:43:02 krowka systemd[1]: Unit user@6.service has alias user@.service.
9 Jun 01 09:43:02 krowka systemd[1]: Unit user-runtime-dir@6.service has alias user-runtime-dir@.service.
13 * Fedora: add an rpmlint check that verifies that all unit files in the RPM are listed in %systemd_post macros.
16 - natively watch for dbus-*.service symlinks (PENDING)
17 - teach dbus to activate all services it finds in /etc/systemd/services/org-*.service
19 * kernel: add device_type = "fb", "fbcon" to class "graphics"
21 * /usr/bin/service should actually show the new command line
23 * 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=
25 * neither pkexec nor sudo initialize environ[] from the PAM environment?
27 * fedora: update policy to declare access mode and ownership of unit files to root:root 0644, and add an rpmlint check for it
29 * register catalog database signature as file magic
31 * zsh shell completion:
32 - <command> <verb> -<TAB> should complete options, but currently does not
33 - systemctl add-wants,add-requires
34 - systemctl reboot --boot-loader-entry=
36 * systemctl status should know about 'systemd-analyze calendar ... --iterations='
37 * If timer has just OnInactiveSec=..., it should fire after a specified time
40 * write blog stories about:
41 - hwdb: what belongs into it, lsusb
42 - enabling dbus services
43 - how to make changes to sysctl and sysfs attributes
45 - how to pass throw-away units to systemd, or dynamically change properties of existing units
46 - testing with Harald's awesome test kit
48 - how to develop against journal browsing APIs
49 - the journal HTTP iface
50 - non-cgroup resource management
51 - dynamic resource management with cgroups
52 - refreshed, longer missions statement
53 - calendar time events
54 - init=/bin/sh vs. "emergency" mode, vs. "rescue" mode, vs. "multi-user" mode, vs. "graphical" mode, and the debug shell
55 - how to create your own target
56 - instantiated apache, dovecot and so on
57 - hooking a script into various stages of shutdown/rearly booot
61 * look for close() vs. close_nointr() vs. close_nointr_nofail()
63 * check for strerror(r) instead of strerror(-r)
67 * set_put(), hashmap_put() return values check. i.e. == 0 does not free()!
69 * use secure_getenv() instead of getenv() where appropriate
71 * link up selected blog stories from man pages and unit files Documentation= fields
75 * rework mount.c and swap.c to follow proper state enumeration/deserialization
76 semantics, like we do for device.c now
78 * get rid of prefix_roota() and similar, only use chase() and related
81 * get rid of basename() and replace by path_extract_filename()
83 * Replace our fstype_is_network() with a call to libmount's mnt_fstype_is_netfs()?
84 Having two lists is not nice, but maybe it's now worth making a dependency on
85 libmount for something so trivial.
87 * drop set_free_free() and switch things over from string_hash_ops to
88 string_hash_ops_free everywhere, so that destruction is implicit rather than
89 explicit. Similar, for other special hashmap/set/ordered_hashmap destructors.
91 * generators sometimes apply C escaping and somethines specifier escaping to
92 paths and similar strings they write out. Sometimes both. We should clean
93 this up, and should probably always apply both, i.e. introduce
94 unit_file_escape() or so, which applies both.
96 Deprecations and removals:
98 * Remove any support for booting without /usr pre-mounted in the initrd entirely.
99 Update INITRD_INTERFACE.md accordingly.
101 * remove cgrouspv1 support EOY 2023. As per
102 https://lists.freedesktop.org/archives/systemd-devel/2022-July/048120.html
103 and then rework cgroupsv2 support around fds, i.e. keep one fd per active
104 unit around, and always operate on that, instead of cgroup fs paths.
106 * drop support for kernels that lack ambient capabilities support (i.e. make
107 4.3 new baseline). Then drop support for "!!" modifier for ExecStart= which
108 is only supported for such old kernels.
110 * drop support for kernels lacking memfd_create() (i.e. make 3.17 new
111 baseline), then drop all pipe() based fallbacks.
113 * drop support for getrandom()-less kernels. (GRND_INSECURE means once kernel
114 5.6 becomes our baseline). See
115 https://github.com/systemd/systemd/pull/24101#issuecomment-1193966468 for
116 details. Maybe before that: at taint-flags/warn about kernels that lack
117 getrandom()/environments where it is blocked.
119 * drop support for LOOP_CONFIGURE-less loopback block devices, once kernel
122 * drop fd_is_mount_point() fallback mess once we can rely on
123 STATX_ATTR_MOUNT_ROOT to exist i.e. kernel baseline 5.8
125 * rework our PID tracking in services and so on, to be strictly based on pidfd,
126 once kernel baseline is 5.13.
128 * Remove /dev/mem ACPI FPDT parsing when /sys/firmware/acpi/fpdt is ubiquitous.
129 That requires distros to enable CONFIG_ACPI_FPDT, and have kernels v5.12 for
130 x86 and v6.2 for arm.
132 * Once baseline is 4.13, remove support for INTERFACE_OLD= checks in "udevadm
133 trigger"'s waiting logic, since we can then rely on uuid-tagged uevents
135 * remove remaining tpm1.2 support from sd-stub
139 * ddi must be listed as block device fstype
141 * measure some string via pcrphase whenever we end up booting into emergency
144 * homed: add a basic form of of secrets management to homed, that stores
145 secrets in $HOME somewhere, is protected by the accounts own authentication
146 mechanisms. Should implement something PKCS#11-like that can be used to
147 implement emulated FIDO2 in unpriv userspace on top (which should happen
148 outside of homed), emulated PKCS11, and libsecrets support. Operate with a
149 2nd key derived from volume key of the user, with which to wrap all
150 keys. maintain keys in kernel keyring if possible.
152 * add ConditionSecurity=stub-measured or so that checks if we are booted with
153 systemd-stub and its measurements
155 * sd-boot should probably measure its configuration file to PCR 5 at boot, as
156 per TCG PC Client Platform Firmware Profile Spec.
158 * use sd-event ratelimit feature optionally for .socket units to "pause" overly
159 busy sockets temporarily. (as a less drastic version of the trigger
162 * similar, add the same for journal stream clients that log too much
164 * systemd-mount sould only consider modern file systems when mounting, similar
167 * new "systemd-pcrlock" component for dealing with PCR4. Design idea:
168 1. define /{etc,usr,var/lib}/pcrlock.d/<component>/<version>.pcrlock
169 2. these files contain list of hashes that will be measured when component is
171 3. each component involved in the boot that is deterministically measured can
172 place one or more of these files in those dirs (shim, sd-boot,
173 sd-stub/UKI, cryptsetup, pcrphase, pcrfs, …)
174 4. since each component has its own dir, with multiple files in them, package
175 such as kernels (of which there can be multiple installed at the same
176 time) can be grouped together: only one of them is measured at a time.
177 5. whenever a new component is added or an old one removed, or the PCR lock
178 shall be relaxed or tightened the systemd-pcrlock tool is invoked.
179 6. tool iterates through all these files, orders them alphabetically by
180 component, then matches them up with current measurements (as per uefi
181 event log), identifying by hash, accepting that the "beginning" of the
182 measurements might not be recognizable.
183 7. Then calculates expected PCR values starting with the "unrecognized
184 head" from the event log, then continuing with all of components
185 defined via the .pcrlock files (but dropping out the "recognized tail"
186 from the uefi event log). (This might mean combinatorial explosion, if
187 there are multiple shims, multiple sd-boot, and so on.)
188 8. Generates a public/private key pair on the TPM
189 9. Generates a counter object in the TPM, with a policy that allows only
190 one-by-one increase with signature policy by the public/private key pair.
191 10. now signs policies of all expected PCR values with the generated keypair,
192 using all combinations of components defined in the .pcrlock files
193 restricting it to the counter + 1.
194 11. locks down the keypair with a signed policy with its own public key
195 12. generates JSON file of all these policies with their signatures, drops
196 them as singleton in ESP
197 13. increases the counter by one.
198 14. after boot sd-stub picks JSON up from ESP, passes it to userspace via
200 15. JSON contained policies can now be used to unlock disk as well as the
201 public/key itself for signing further policies, as well as increment for
203 16. whenever any of the components above is added/removed new JSON file with
204 signatures for counter + 1 is generated, dropped in ESP, then counter
205 increased. (i.e. this means the "recognized tail" of the event log is
206 deterministically swapped out)
207 17. when firmware update is expected, relaxed signed policy is generated for
208 next boot only valid if counter is increased (this means the
209 "unrecognized head" for the event log can change without losing access)
210 18. on every boot checks if releaxed policy is in effect, if so, new strict
211 policy is generated and counter increased.
212 Net result: Removes downgrade attack surface + Locks OS to firmware + Allows
213 downgrades within bounds
215 * add another PE section ".fname" or so that encodes the intended filename for
216 PE file, and validate that when loading add-ons and similar before using
217 it. This is particularly relevant when we load multiple add-ons and want to
218 sort them to apply them in a define order. The order should not be under
219 control of the attacker.
221 * also include packaging metadata (á la
222 https://systemd.io/ELF_PACKAGE_METADATA/) in our UEFI PE binaries, using the
225 * make "bootctl install" + "bootctl update" useful for installing shim too. For
226 that introduce new dir /usr/lib/systemd/efi/extra/ which we copy mostly 1:1
227 into the ESP at install time. Then make the logic smart enough so that we
228 don't overwrite bootx64.efi with our own if the extra tree already contains
229 one. Also, follow symlinks when copying, so that shim rpm can symlink their
230 stuff into our dir (which is safe since the target ESP is generally VFAT and
231 thus does not have symlinks anyway). Later, teach the update logic to look at
232 the ELF package metadata (which we also should include in all PE files, see
233 above) for version info in all *.EFI files, and use it to only update if
236 * in sd-stub: optionally add support for a new PE section .keyring or so that
237 contains additional certificates to include in the Mok keyring, extending
238 what shim might have placed there. why? let's say I use "ukify" to build +
239 sign my own fedora-based UKIs, and only enroll my personal lennart key via
240 shim. Then, I want to include the fedora keyring in it, so that kmods work.
241 But I might not want to enroll the fedora key in shim, because this would
242 also mean that the key would be in effect whenever I boot an archlinux UKI
243 built the same way, signed with the same lennart key.
245 * resolved: take possession of some IPv6 ULA address (let's say
246 fd00:5353:5353:5353:5353:5353:5353:5353), and listen on port 53 on it for the
247 local stubs, so that we can make the stub available via ipv6 too.
249 * introduce a .microcode PE section for sd-stub which we'll pass as first initrd
250 to the kernel which will then upload it to the CPU. This should be distinct
251 from .initrd to guarantee right ordering. also, and maybe more importantly
252 support .microcode in PE add-ons, so that a microcode update can be shipped
253 independently of any kernel.
255 * Maybe add SwitchRootEx() as new bus call that takes env vars to set for new
256 PID 1 as argument. When adding SwitchRootEx() we should maybe also add a
257 flags param that allows disabling and enabling whether serialization is
258 requested during switch root.
260 * introduce a .acpitable section for early ACPI table override
262 * add proper .osrel matching for PE addons. i.e. refuse applying an addon
263 intended for a different OS. Take inspiration from how confext/sysext are
266 * use different sbat for sd-boot and sd-stub (so that people can revoke one
269 * in ukify merge sbat info from kernel (if it has any, upstream kernels so far
270 dont), of sd-stub and data supplied by user. Then measure sbat too in
273 * figure out what to do about credentials sealed to PCRs in kexec + soft-reboot
274 scenarios. Maybe insist sealing is done additionally against some keypair in
275 the TPM to which access is updated on each boot, for the next, or so?
277 * logind: when logging in, always take an fd to the home dir, to keep the dir
278 busy, so that autofs release can never happen. (this is generally a good
279 idea, and specifically works around the fact the autofs ignores busy by mount
282 * mount most file systems with a restrictive uidmap. e.g. mount /usr/ with a
283 uidmap that blocks out anything outside 0…1000 (i.e. system users) and similar.
285 * mount the root fs with MS_NOSUID by default, and then mount /usr/ without
286 both so that suid executables can only be placed there. Do this already in
287 the initrd. If /usr/ is not split out create a bind mount automatically.
289 * fix our various hwdb lookup keys to end with ":" again. The original idea was
290 that hwdb patterns can match arbitrary fields with expressions like
291 "*:foobar:*", to wildcard match both the start and the end of the string.
292 This only works safely for later extensions of the string if the strings
293 always end in a colon. This requires updating our udev rules, as well as
294 checking if the various hwdb files are fine with that.
296 * mount /tmp/ and /var/tmp with a uidmap applied that blocks out "nobody" user
297 among other things such as dynamic uid ranges for containers and so on. That
298 way no one can create files there with these uids and we enforce they are only
299 used transiently, never persistently.
301 * rework loopback support in fstab: when "loop" option is used, then
302 instantiate a new systemd-loop@.service for the source path, set the
303 lo_file_name field for it to something recognizable derived from the fstab
304 line, and then generate a mount unit for it using a udev generated symlink
305 based on lo_file_name.
307 * remove tomoyo support, it's obsolete and unmaintained apparently
309 * In .socket units, add ConnectStream=, ConnectDatagram=,
310 ConnectSequentialPacket= that create a socket, and then *connect to* rather than
311 listen on some socket. Then, add a new setting WriteData= that takes some
312 base64 data that systemd will write into the socket early on. This can then
313 be used to create connections to arbitrary services and issue requests into
314 them, as long as the data is static. This can then be combined with the
315 aforementioned journald subscription varlink service, to enable
316 activation-by-message id and similar.
318 * .service with invalid Sockets= starts successfully.
320 * landlock: lock down RuntimeDirectory= via landlock, so that services lose
321 ability to write anywehere else below /run/. Similar for
322 StateDirectory=. Benefit would be clear delegation via unit files: services
323 get the directories they get, and nothing else even if they wanted to.
325 * landlock: for unprivileged systemd (i.e. systemd --user), use landlock to
326 implement ProtectSystem=, ProtectHome= and so on. Landlock does not require
327 privs, and we can implement pretty similar behaviour. Also, maybe add a mode
328 where ProtectSystem= combined with an explicit PrivateMounts=no could request
329 similar behaviour for system services, too.
331 * Add systemd-mount@.service which is instantiated for a block device and
332 invokes systemd-mount and exits. This is then useful to use in
333 ENV{SYSTEMD_WANTS} in udev rules, and a bit prettier than using RUN+=
335 * udevd: extend memory pressure logic: also kill any idle worker processes
337 * SIGRTMIN+18 and memory pressure handling should still be added to: hostnamed,
338 localed, oomd, timedated.
340 * in order to make binding to PCR 4 realistic:
341 - generate one keypair "U" and store it in a tpm2 nvindex.
342 - Generate another keypair "P" and store it in a second tpm2 nvindex.
343 - allocate a persistent counter object "C" in the tpm2
344 - Enroll all user objects (i.e. luks volumes, creds, …) to a tpm2 policy
346 - Lock both U and P down with a tpm2 policy signed by P (yes, P can only be
347 used if a signature by P itself can be provided)
348 - For regular reboots generate a signature for a restrictive PCR4 + counter C
349 based policy with key P. Place signature in EFI var, so it can be found on
351 - For reboots where a firmware update is expected generate a signature with a
352 more open policy against just counter C. Place signature in same EFI var.
353 - Increase C whenever switching between these two signature types.
354 - During early boot, use the signature from the EFI var to unlock U and P.
355 Use it to generate a signature for unlocking user objects given the current
356 PCR 4 value, store that away into /run somewhere, for user during the whole
358 - When booting up automatically update the mentioned efi var so that it
359 contains the restrictive signature. But also generate a signature ahead of
360 time that could be used in case during the current boot we later detect we might
361 need to reboot for a firmware update. Store that in /run somewhere, so that
362 it can be placed in the EFI var, if needed.
364 * repart/gpt-auto/DDIs: maybe introduce a concept of "extension" partitions,
365 that have a new type uuid and can "extend" earlier partitions, to work around
366 the fact that systemd-repart can only grow the last partition defined. During
367 activation we'd simply set up a dm-linear mapping to merge them again. A
368 partition that is to be extended would just set a bit in the partition flags
369 field to indicate that there's another extension partition to look for. The
370 identifying UUID of the extension partition would be hashed in counter mode
371 from the uuid of the original partition it extends. Inspiration for this is
372 the "dynamic partitions" concept of new Android. This would be a minimalistic
373 concept of a volume manager, with the extents it manages being exposes as GPT
374 partitions. I a partition is extended multiple times they should probably
375 grow exponentially in size to ensure O(log(n)) time for finding them on
378 * split out execute.c into new "systemd-executor" binary. Then make PID 1 fork
379 that off via vfork(), and then let that executor do the hard work. Ultimately
380 the executor then gets replaced by the real binary sooner or later. Reason:
381 currently the intermediary "stub" process is a CoW trap that doubles memory
382 usage of PID 1 on each service start. Also, strictly speaking we are not
383 allowed to do NSS from the stub process yet we do anyway. Next steps would
384 then be maybe use CLONE_INTO_CGROUP for the executor, given that we don't
385 need glibc anymore in the stub process then. Then, switch nspawn to just be a
386 frontend for this too, so that we have to ways into the executor: via unit
387 files/dbus/varlin through PID1 and via cmdline/OCI through nspawn.
389 * sd-stub: detect if we are running with uefi console output on serial, and if so
390 automatically add console= to kernel cmdline matching the same port.
392 * add a utility that can be used with the kernel's
393 CONFIG_STATIC_USERMODEHELPER_PATH and then handles them within pid1 so that
394 security, resource management and cgroup settings can be enforced properly
395 for all umh processes.
397 * systemd-shutdown: keep sending sd_notify() status updates immediately before
398 going down, in particular include the "reboot param" string.
400 * homed: when resizing an fs don't sync identity beforehand there might simply
401 not be enough disk space for that. try to be defensive and sync only after
404 * homed: if for some reason the partition ended up being much smaller than
405 whole disk, recover from that, and grow it again.
407 * timesyncd: when saving/restoring clock try to take boot time into account.
408 Specifically, along with the saved clock, store the current boot ID. When
409 starting, check if the boot id matches. If so, don't do anything (we are on
410 the same boot and clock just kept running anyway). If not, then read
411 CLOCK_BOOTTIME (which started at boot), and add it to the saved clock
412 timestamp, to compensate for the time we spent booting. If EFI timestamps are
413 available, also include that in the calculation. With this we'll then only
414 miss the time spent during shutdown after timesync stopped and before the
415 system actually reset.
417 * systemd-stub: maybe store a "boot counter" in the ESP, and pass it down to
418 userspace to allow ordering boots (for example in journalctl). The counter
419 would be monotonically increased on every boot.
421 * pam_systemd_home: add module parameter to control whether to only accept
422 only password or only pcks11/fido2 auth, and then use this to hook nicely
423 into two of the three PAM stacks gdm provides.
424 See discussion at https://github.com/authselect/authselect/pull/311
426 * sd-boot: make boot loader spec type #1 accept http urls in "linux"
427 lines. Then, do the uefi http dance to download kernels and boot them. This
428 is then useful for network boot, by embdedding a cpio with type #1 snippets
429 in sd-boot, which reference remote kernels.
431 * maybe prohibit setuid() to the nobody user, to lock things down, via seccomp.
432 the nobody is not a user any code should run under, ever, as that user would
433 possibly get a lot of access to resources it really shouldn't be getting
434 access to due to the userns + nfs semantics of the user. Alternatively: use
435 the seccomp log action, and allow it.
437 * sd-boot: add a new PE section .bls or so that carries a cpio with additional
438 boot loader entries (both type1 and type2). Then when initializing, find this
439 section, iterate through it and populate menu with it. cpio is simple enough
440 to make a parser for this reasonably robust. use same path structures as in
441 the ESP. Similar add one for signature key drop-ins.
443 * sd-boot: also allow passing in the cpio as in the previous item via SMBIOS
445 * add a new EFI tool "sd-fetch" or so. It looks in a PE section ".url" for an
446 URL, then downloads the file from it using UEFI HTTP APIs, and executes it.
447 Usecase: provide a minimal ESP with sd-boot and a couple of these sd-fetch
448 binaries in place of UKIs, and download them on-the-fly.
450 * maybe: systemd-loop-generator that sets up loopback devices if requested via kernel
451 cmdline. usecase: include encrypted/verity root fs in UKI.
453 * systemd-gpt-auto-generator: add kernel cmdline option to override block
454 device to dissect. also support dissecting a regular file. useccase: include
455 encrypted/verity root fs in UKI.
457 * sd-stub: add ".bootcfg" section for kernel bootconfig data (as per
458 https://docs.kernel.org/admin-guide/bootconfig.html)
460 * tpm2: add (optional) support for generating a local signing key from PCR 15
461 state. use private key part to sign PCR 7+14 policies. stash signatures for
462 expected PCR7+14 policies in EFI var. use public key part in disk encryption.
463 generate new sigs whenever db/dbx/mok/mokx gets updated. that way we can
464 securely bind against SecureBoot/shim state, without having to renroll
465 everything on each update (but we still have to generate one sig on each
466 update, but that should be robust/idempotent). needs rollback protection, as
469 * Lennart: big blog story about DDIs
471 * Lennart: big blog story about building initrds
473 * Lennart: big blog story about "why systemd-boot"
475 * bpf: see if we can use BPF to solve the syslog message cgroup source problem:
476 one idea would be to patch source sockaddr of all AF_UNIX/SOCK_DGRAM to
477 implicitly contain the source cgroup id. Another idea would be to patch
478 sendto()/connect()/sendmsg() sockaddr on-the-fly to use a different target
481 * bpf: see if we can address opportunistic inode sharing of immutable fs images
482 with BPF. i.e. if bpf gives us power to hook into openat() and return a
483 different inode than is requested for which we however it has same contents
484 then we can use that to implement opportunistic inode sharing among DDIs:
485 make all DDIs ship xattr on all reg files with a SHA256 hash. Then, also
486 dictate that DDIs should come with a top-level subdir where all reg files are
487 linked into by their SHA256 sum. Then, whenever an inode is opened with the
488 xattr set, check bpf table to find dirs with hashes for other prior DDIs and
489 try to use inode from there.
491 * extend the verity signature partition to permit multiple signatures for the
492 same root hash, so that people can sign a single image with multiple keys.
494 * consider adding a new partition type, just for /opt/ for usage in system
497 * gpt-auto-discovery: also use the pkcs7 signature stuff, and pass signature to
498 kernel. So far we only did this for the various --image= switches, but not
499 for the root fs or /usr/.
501 * dissection policy should enforce that unlocking can only take place by
502 certain means, i.e. only via pw, only via tpm2, or only via fido, or a
505 * make the systemd-repart "seed" value provisionable via credentials, so that
506 confidential computing environments can set it and deterministically
507 enforce the uuids for partitions created, so that they can calculate PCR 15
510 * systemd-repart: also derive the volume key from the seed value, for the
511 aforementioned purpose.
513 * in the initrd: derive the default machine ID to pass to the host PID 1 via
514 $machine_id from the same seed credential.
516 * Add systemd-sysupdate-initrd.service or so that runs systemd-sysupdate in the
517 initrd to bootstrap the initrd to populate the initial partitions. Some things
519 - Should it run on firstboot or on every boot?
520 - If run on every boot, should it use the sysupdate config from the host on
523 * provide an API (probably IPC) to apps to encrypt/decrypt
524 credentials. usecase: allow bluez bluetooth daemon to pass pairings to initrd
525 that way, without shelling out to our tools.
527 * revisit default PCR bindings in cryptenroll and systemd-creds. Currently they
528 use PCR 7 which should contain secureboot state db/dbx. Which sounded like a
529 safe bet, given that it should change only on policy changes, and not
530 software updates. But that's wrong. Recent fwupd (rightfully) contains code
531 for updating the dbx denylist. This means even without any active policy
532 change PCR 7 might change. Hence, better idea might be in systemd-creds to
533 default to PCR 15 at least if sd-stub is used (i.e. bind to system identity),
534 and in cryptsetup simply the empty list? Also, PCR 14 almost certainly should
535 be included as much as PCR 7 (as it contains shim's policy, which is
536 certainly as relevant as PCR 7 on many systems)
538 * To mimic the new tpm2-measure-pcr= crypttab option add the same to veritytab
539 (measuring the root hash) and integritytab (measuring the HMAC key if one is
542 * We should start measuring all services, containers, and system extensions we
543 activate. probably into PCR 13. i.e. add --tpm2-measure-pcr= or so to
544 systemd-nspawn, and MeasurePCR= to unit files. Should contain a measurement
545 of the activated configuration and the image that is being activated (in case
546 verity is used, hash of the root hash).
548 * whenever we measure something into a TPM PCR from userspace, write a record in
549 TCG's "Canonical Event Log" format to some file, so that we can reason about
550 how PCR values we manage came to
551 be. https://trustedcomputinggroup.org/resource/canonical-event-log-format/
553 * bootspec: permit graceful "update" from type #2 to type #1. If both a type #1
554 and a type #2 entry exist under otherwise the exact same name, then use the
555 type #1 entry, and ignore the type #2 entry. This way, people can "upgrade"
556 from the UKI with all parameters baked in to a Type #1 .conf file with manual
557 parametrization, if needed. This matches our usual rule that admin config
558 should win over vendor defaults.
560 * write a "search path" spec, that documents the prefixes to search in
561 (i.e. the usual /etc/, /run/, /usr/lib/ dance, potentially /usr/etc/), how to
562 sort found entries, how masking works and overriding.
564 * automatic boot assessment: add one more default success check that just waits
565 for a bit after boot, and blesses the boot if the system stayed up that long.
567 * implement concept of "versioned" resources inside a dir, and write a spec for
568 it. Make all tools in systemd, in particular
569 RootImage=/RootDirectory=/--image=/--directory= implement this. Idea:
570 directories ending in ".v/" indicate a directory with versioned resources in
571 them. Versioned resources inside a .v dir are always named in the pattern
572 <prefix>_<version>[+<tries-left>[-<tries-done>]].<suffix>
574 * add support for using this .v/ logic on the root fs itself: in the initrd,
575 after mounting the rootfs, look for root-<arch>.v/ in the root fs, and then
576 apply the logic, moving the switch root logic there.
578 * systemd-repart: add support for generating ISO9660 images
580 * systemd-repart: in addition to the existing "factory reset" mode (which
581 simply empties existing partitions marked for that). add a mode where
582 partitions marked for it are entirely removed. Usecase: remove secondary OS
583 copy, and redundant partitions entirely, and recreate them anew.
585 * systemd-boot: maybe add support for collapsing menu entries of the same OS
586 into one item that can be opened (like in a "tree view" UI element) or
587 collapsed. If only a single OS is installed, disable this mode, but if
588 multiple OSes are installed might make sense to default to it, so that user
589 is not immediately bombarded with a multitude of Linux kernel versions but
590 only one for each OS.
592 * systemd-repart: if the GPT *disk* UUID (i.e. the one global for the entire
593 disk) is set to all FFFFF then use this as trigger for factory reset, in
594 addition to the existing mechanisms via EFI variables and kernel command
595 line. Benefit: works also on non-EFI systems, and can be requested on one
598 * figure out a sane way when building UKIs how to extract SBAT data from inner
599 kernel, extend it with component info, and add to outer kernel.
601 * systemd-sysupdate: make transport pluggable, so people can plug casync or
602 similar behind it, instead of http.
604 * systemd-tmpfiles: add concept for conditionalizing lines on factory reset
605 boot, or on first boot.
607 * in UKIs: add way to define allowlist of additional words that can be added to
608 the kernel cmdline even in SecureBoot mode
610 * we probably needs .pcrpkeyrd or so as additional PE section in UKIs,
611 which contains a separate public key for PCR values that only apply in the
612 initrd, i.e. in the boot phase "enter-initrd". Then, consumers in userspace
613 can easily bind resources to just the initrd. Similar, maybe one more for
614 "enter-initrd:leave-initrd" for resources that shall be accessible only
615 before unprivileged user code is allowed. (we only need this for .pcrpkey,
616 not for .pcrsig, since the latter is a list of signatures anyway). With that,
617 when you enroll a LUKS volume or similar, pick either the .pcrkey (for
618 coverage through all phases of the boot, but excluding shutdown), the
619 .pcrpkeyrd (for coverage in the initrd only) and .pcrpkeybt (for coverage
620 until users are allowed to log in).
622 * Once the root fs LUKS volume key is measured into PCR 15, default to binding
623 credentials to PCR 15 in "systemd-creds"
625 * add support for asymmetric LUKS2 TPM based encryption. i.e. allow preparing
626 an encrypted image on some host given a public key belonging to a specific
627 other host, so that only hosts possessing the private key in the TPM2 chip
628 can decrypt the volume key and activate the volume. Usecase: systemd-confext
629 for a central orchestrator to generate confext images securely that can only
630 be activated on one specific host (which can be used for installing a bunch
631 of creds in /etc/credstore/ for example). Extending on this: allow binding
632 LUKS2 TPM based encryption also to the TPM2 internal clock. Net result:
633 prepare a confext image that can only be activated on a specific host that
634 runs a specific software in a specific time window. confext would be
635 automatically invalidated outside of it.
637 * maybe add a "systemd-report" tool, that generates a TPM2-backed "report" of
638 current system state, i.e. a combination of PCR information, local system
639 time and TPM clock, running services, recent high-priority log
640 messages/coredumps, system load/PSI, signed by the local TPM chip, to form an
641 enhanced remote attestation quote. Usecase: a simple orchestrator could use
642 this: have the report tool upload these reports every 3min somewhere. Then
643 have the orchestrator collect these reports centrally over a 3min time
644 window, and use them to determine what which node should now start/stop what,
645 and generate a small confext for each node, that uses Uphold= to pin services
646 on each node. The confext would be encrypted using the asymmetric encryption
647 proposed above, so that it can only be activated on the specific host, if the
648 software is in a good state, and within a specific time frame. Then run a
649 loop on each node that sends report to orchestrator and then sysupdate to
650 update confext. Orchestrator would be stateless, i.e. operate on desired
651 config and collected reports in the last 3min time window only, and thus can
652 be trivially scaled up since all instances of the orchestrator should come to
653 the same conclusions given the same inputs of reports/desired workload info.
654 Could also be used to deliver Wireguard secrets and thus to clients, thus
655 permitting zero-trust networking: secrets are rolled over via confext updates,
656 and via the time window TPM logic invalidated if node doesn't keep itself
657 updated, or becomes corrupted in some way.
659 * in the initrd, once the rootfs encryption key has been measured to PCR 15,
660 derive default machine ID to use from it, and pass it to host PID 1.
662 * tree-wide: convert as much as possible over to use sd_event_set_signal_exit(), instead
663 of manually hooking into SIGINT/SIGTERM
665 * tree-wide: convert as much as possible over to SD_EVENT_SIGNAL_PROCMASK
666 instead of manual blocking.
668 * sd-boot: for each installed OS, grey out older entries (i.e. all but the
669 newest), to indicate they are obsolete
671 * automatically propagate LUKS password credential into cryptsetup from host
672 (i.e. SMBIOS type #11, …), so that one can unlock LUKS via VM hypervisor
675 * add ability to path_is_valid() to classify paths that refer to a dir from
676 those which may refer to anything, and use that in various places to filter
677 early. i.e. stuff ending in "/", "/." and "/.." definitely refers to a
678 directory, and paths ending that way can be refused early in many contexts.
680 * systemd-measure: allow operating with PEM certificates in addition to PEM
681 public keys when signing PCR values. SecureBoot and our Verity signatures
682 operate with certificates already, hence I guess we should also just deal for
683 convencience with certificates for the PCR stuff too.
685 * systemd-measure: add --pcrpkey-auto as an alternative to --pcrpkey=, where it
686 would just use the same public key specified with --public-key= (or the one
687 automatically derived from --private-key=).
689 * push people to use ".sysext.raw" as suffix for sysext DDIs (DDI =
690 discoverable disk images, i.e. the new name for gpt disk images following the
691 discoverable disk spec). [Also: just ".sysext/" for directory-based sysext]
693 * Add "purpose" flag to partition flags in discoverable partition spec that
694 indicate if partition is intended for sysext, for portable service, for
695 booting and so on. Then, when dissecting DDI allow specifying a purpose to
696 use as additional search condition. Usecase: images that combined a sysext
697 partition with a portable service partition in one.
699 * On boot, auto-generate an asymmetric key pair from the TPM,
700 and use it for validating DDIs and credentials. Maybe upload it to the kernel
701 keyring, so that the kernel does this validation for us for verity and kernel
704 * for systemd-confext: add a tool that can generate suitable DDIs with verity +
705 sig using squashfs-tools-ng's library. Maybe just systemd-repart called under
706 a new name with a built-in config?
708 * lock down acceptable encrypted credentials at boot, via simple allowlist,
709 maybe on kernel command line:
710 systemd.import_encrypted_creds=foobar.waldo,tmpfiles.extra to protect locked
711 down kernels from credentials generated on the host with a weak kernel
713 * Add support for extra verity configuration options to systemd-repart (FEC,
716 * chase(): take inspiration from path_extract_filename() and return
717 O_DIRECTORY if input path contains trailing slash.
719 * chase(): refuse resolution if trailing slash is specified on input,
720 but final node is not a directory
722 * document in boot loader spec that symlinks in XBOOTLDR/ESP are not OK even if
725 * measure credentials picked up from SMBIOS to some suitable PCR
727 * measure GPT and LUKS headers somewhere when we use them (i.e. in
728 systemd-gpt-auto-generator/systemd-repart and in systemd-cryptsetup?)
730 * pick up creds from EFI vars
732 * Add and pickup tpm2 metadata for creds structure.
734 * sd-boot: we probably should include all BootXY EFI variable defined boot
735 entries in our menu, and then suppress ourselves. Benefit: instant
736 compatibility with all other OSes which register things there, in particular
737 on other disks. Always boot into them via NextBoot EFI variable, to not
740 * systemd-measure tool:
741 - pre-calculate PCR 12 (command line) + PCR 13 (sysext) the same way we can precalculate PCR 11
743 * in sd-boot: load EFI drivers from a new PE section. That way, one can have a
744 "supercharged" sd-boot binary, that could carry ext4 drivers built-in.
746 * sd-bus: document that sd_bus_process() only returns messages that non of the
747 filters/handlers installed on the connection took possession of.
749 * sd-device: add an API for acquiring list of child devices, given a device
750 objects (i.e. all child dirents that dirs or symlinks to dirs)
752 * sd-device: maybe pin the sysfs dir with an fd, during the entire runtime of
753 an sd_device, then always work based on that.
755 * add small wrapper around qemu that implements sd_notify/AF_VSOCK + machined and
756 maybe some other stuff and boots it. Should implement command line roughly
757 equivalent to nspawn's. Maybe be called "systemd-vmspawn". Should imply good
758 settings, i.e. RNG + HyperV enlightenments. Should also result in swtpm
759 instance, plus virtiofsd instances. Translate credentials into smbios type
760 11 strings. Correctly translate SIGTERM into ACPI shutdown events.
761 Listen to logind suspend events and turn these into suspend key pressed +
764 * maybe add new flags to gpt partition tables for rootfs and usrfs indicating
765 purpose, i.e. whether something is supposed to be bootable in a VM, on
766 baremetal, on an nspawn-style container, if it is a portable service image,
767 or a sysext for initrd, for host os, or for portable container. Then hook
768 portabled/… up to udev to watch block devices coming up with the flags set, and
771 * sd-boot should look for information what to boot in SMBIOS, too, so that VM
772 managers can tell sd-boot what to boot into and suchlike
774 * add "systemd-sysext identify" verb, that you can point on any file in /usr/
775 and that determines from which overlayfs layer it originates, which image, and with
778 * systemd-creds: extend encryption logic to support asymmetric
779 encryption/authentication. Idea: add new verb "systemd-creds public-key"
780 which generates a priv/pub key pair on the TPM2 and stores the priv key
781 locally in /var. It then outputs a certificate for the pub part to stdout.
782 This can then be copied/taken elsewhere, and can be used for encrypting creds
783 that only the host on its specific hw can decrypt. Then, support a drop-in
784 dir with certificates that can be used to authenticate credentials. Flow of
785 operations is then this: build image with owner certificate, then after
786 boot up issue "systemd-creds public-key" to acquire pubkey of the machine.
787 Then, when passing data to the machine, sign with privkey belonging to one of
788 the dropped in certs and encrypted with machine pubkey, and pass to machine.
789 Machine is then able to authenticate you, and confidentiality is guaranteed.
791 * building on top of the above, the pub/priv key pair generated on the TPM2
792 should probably also one you can use to get a remote attestation quote.
794 * Process credentials in:
795 • networkd/udevd: add a way to define additional .link, .network, .netdev files
796 via the credentials logic.
797 • crypttab-generator: allow defining additional crypttab-like volumes via
798 credentials (similar: verity-generator, integrity-generator). Use
799 fstab-generator logic as inspiration.
800 • run-generator: allow defining additional commands to run via a credential
801 • resolved: allow defining additional /etc/hosts entries via a credential (it
802 might make sense to then synthesize a new combined /etc/hosts file in /run
803 and bind mount it on /etc/hosts for other clients that want to read it.
804 • repart: allow defining additional partitions via credential
805 • timesyncd: pick NTP server info from credential
806 • portabled: read a credential "portable.extra" or so, that takes a list of
807 file system paths to enable on start.
808 • make systemd-fstab-generator look for a system credential encoding root= or
810 • systemd-homed: when initializing, look for a credential
811 systemd.homed.register or so with JSON user records to automatically
812 register if not registered yet. Usecase: deploy a system, and add an
813 account one can directly log into.
814 • in gpt-auto-generator: check partition uuids against such uuids supplied via
815 sd-stub credentials. That way, we can support parallel OS installations with
818 * define a JSON format for units, separating out unit definitions from unit
819 runtime state. Then, expose it:
821 1. Add Describe() method to Unit D-Bus object that returns a JSON object
823 2. Expose this natively via Varlink, in similar style
824 3. Use it when invoking binaries (i.e. make PID 1 fork off systemd-executor
825 binary which reads the JSON definition and runs it), to address the cow
826 trap issue and the fact that NSS is actually forbidden in
827 forked-but-not-exec'ed children
828 4. Add varlink API to run transient units based on provided JSON definitions
830 * Add SUPPORT_END_URL= field to os-release with more *actionable* information
831 what to do if support ended
833 * pam_systemd: on interactive logins, maybe show SUPPORT_END information at
834 login time, à la motd
836 * sd-boot: instead of unconditionally deriving the ESP to search boot loader
837 spec entries in from the paths of sd-boot binary, let's optionally allow it
838 to be configured on sd-boot cmdline + efi var. Usecase: embed sd-boot in the
839 UEFI firmware (for example, ovmf supports that via qemu cmdline option), and
840 use it to load stuff from the ESP.
842 * mount /var/ from initrd, so that we can apply sysext and stuff before the
843 initrd transition. Specifically:
844 1. There should be a var= kernel cmdline option, matching root= and usr=
845 2. systemd-gpt-auto-generator should auto-mount /var if it finds it on disk
846 3. mount.x-initrd mount option in fstab should be implied for /var
848 * implement varlink introspection
850 * make persistent restarts easier by adding a new setting OpenPersistentFile=
851 or so, which allows opening one or more files that is "persistent" across
852 service restarts, hot reboot, cold reboots (depending on configuration): the
853 files are created empty on first invocation, and on subsequent invocations
854 the files are reboot. The files would be backed by tmpfs, pmem or /var
855 depending on desired level of persistency.
857 * sd-event: add ability to "chain" event sources. Specifically, add a call
858 sd_event_source_chain(x, y), which will automatically enable event source y
859 in oneshot mode once x is triggered. Use case: in src/core/mount.c implement
860 the /proc/self/mountinfo rescan on SIGCHLD with this: whenever a SIGCHLD is
861 seen, trigger the rescan defer event source automatically, and allow it to be
862 dispatched *before* the SIGCHLD is handled (based on priorities). Benefit:
863 dispatch order is strictly controlled by priorities again. (next step: chain
864 event sources to the ratelimit being over)
866 * if we fork of a service with StandardOutput=journal, and it forks off a
867 subprocess that quickly dies, we might not be able to identify the cgroup it
868 comes from, but we can still derive that from the stdin socket its output
869 came from. We apparently don't do that right now.
871 * add ability to set hostname with suffix derived from machine id at boot
873 * add PR_SET_DUMPABLE service setting
875 * homed/userdb: maybe define a "companion" dir for home directories where apps
876 can safely put privileged stuff in. Would not be writable by the user, but
877 still conceptually belong to the user. Would be included in user's quota if
878 possible, even if files are not owned by UID of user. Usecase: container
879 images that owned by arbitrary UIDs, and are owned/managed by the users, but
880 are not directly belonging to the user's UID. Goal: we shouldn't place more
881 privileged dirs inside of unprivileged dirs, and thus containers really
882 should not be placed inside of traditional UNIX home dirs (which are owned by
883 users themselves) but somewhere else, that is separate, but still close
884 by. Inform user code about path to this companion dir via env var, so that
885 container managers find it. the ~/.identity file is also a candidate for a
886 file to move there, since it is managed by privileged code (i.e. homed) and
887 not unprivileged code.
889 * given that /etc/ssh/ssh_config.d/ is a thing now, ship a drop-in for that
890 that hooks up userdbctl ssh-key stuff.
892 * maybe add support for binding and connecting AF_UNIX sockets in the file
893 system outside of the 108ch limit. When connecting, open O_PATH fd to socket
894 inode first, then connect to /proc/self/fd/XYZ. When binding, create symlink
895 to target dir in /tmp, and bind through it.
897 * add a proper concept of a "developer" mode, i.e. where cryptographic
898 protections of the root OS are weakened after interactive confirmation, to
899 allow hackers to allow their own stuff. idea: allow entering developer mode
900 only via explicit choice in boot menu: i.e. add explicit boot menu item for
901 it. When developer mode is entered, generate a key pair in the TPM2, and add
902 the public part of it automatically to keychain of valid code signature keys
903 on subsequent boots. Then provide a tool to sign code with the key in the
904 TPM2. Ensure that boot menu item is the only way to enter developer mode, by
905 binding it to locality/PCRs so that keys cannot be generated otherwise.
907 * services: add support for cryptographically unlocking per-service directories
908 via TPM2. Specifically, for StateDirectory= (and related dirs) use fscrypt to
909 set up the directory so that it can only be accessed if host and app are in
912 * TPM2: extend unlock policy to protect against version downgrades in signed
913 policies: policy probably must take some nvram based generation counter into
914 account that can only monotonically increase and can be used to invalidate
915 old PCR signatures. Otherwise people could downgrade to old signed PCR sets
918 * update HACKING.md to suggest developing systemd with the ideas from:
919 https://0pointer.net/blog/testing-my-system-code-in-usr-without-modifying-usr.html
920 https://0pointer.net/blog/running-an-container-off-the-host-usr.html
922 * sd-event: compat wd reuse in inotify code: keep a set of removed watch
923 descriptors, and clear this set piecemeal when we see the IN_IGNORED event
924 for it, or when read() returns EAGAIN or on IN_Q_OVERFLOW. Then, whenever we
925 see an inotify wd event check against this set, and if it is contained ignore
926 the event. (to be fully correct this would have to count the occurrences, in
927 case the same wd is reused multiple times before we start processing
930 * for vendor-built signed initrds:
931 - kernel-install should be able to install encrypted creds automatically for
932 machine id, root pw, rootfs uuid, resume partition uuid, and place next to
933 EFI kernel, for sd-stub to pick them up. These creds should be locked to
934 the TPM, and bind to the right PCR the kernel is measured to.
935 - kernel-install should be able to pick up initrd sysexts automatically and
936 place them next to EFI kernel, for sd-stub to pick them up.
937 - systemd-fstab-generator should look for rootfs device to mount in creds
938 - systemd-resume-generator should look for resume partition uuid in creds
939 - sd-stub: automatically pick up microcode from ESP (/loader/microcode/*)
940 and synthesize initrd from it, and measure it. Signing is not necessary, as
941 microcode does that on its own. Pass as first initrd to kernel.
943 * Maybe extend the service protocol to support handling of some specific SIGRT
944 signal for setting service log level, that carries the level via the
945 sigqueue() data parameter. Enable this via unit file setting.
947 * sd_notify/vsock: maybe support binding to AF_VSOCK in Type=notify services,
948 then passing $NOTIFY_SOCKET and $NOTIFY_GUESTCID with PID1's cid (typically
949 fixed to "2", i.e. the official host cid) and the expected guest cid, for the
950 two sides of the channel. The latter env var could then be used in an
951 appropriate qemu cmdline. That way qemu payloads could talk sd_notify()
952 directly to host service manager.
954 * sd-device has an API to create an sd_device object from a device id, but has
955 no api to query the device id
957 * sd-device should return the devnum type (i.e. 'b' or 'c') via some API for an
958 sd_device object, so that data passed into sd_device_new_from_devnum() can
961 * sd-event: optionally, if per-event source rate limit is hit, downgrade
962 priority, but leave enabled, and once ratelimit window is over, upgrade
963 priority again. That way we can combat event source starvation without
964 stopping processing events from one source entirely.
966 * sd-event: similar to existing inotify support add fanotify support (given
967 that apparently new features in this area are only going to be added to the
970 * sd-event: add 1st class event source for clock changes
972 * sd-event: add 1st class event source for timezone changes
974 * support uefi/http boots with sd-boot: instead of looking for dropin files in
975 /loader/entries/ dir, look for a file /loader/entries/SHA256SUMS and use that
976 as directory manifest. The file would be a standard directory listing as
977 generated by GNU sha256sums.
979 * sd-boot: maybe add support for embedding the various auxiliary resources we
980 look for right in the sd-boot binary. i.e. take inspiration from sd-stub
981 logic: allow combining sd-boot via ukify with kernels to enumerate, .conf
982 files, drivers, keys to enroll and so on. Then, add whatever we find that way
983 to the menu. Usecase: allow building a single PE image you can boot into via
986 * maybe add a new UEFI stub binary "sd-http". It works similar to sd-stub, but
987 all it does is download a file from a http server, and execute it, after
988 optionally checking its hash sum. idea would be: combine this "sd-http" stub
989 binary with some minimal info about a URL + hash sum, plus .osrel data, and
990 drop it into the unified kernel dir in the ESP. And bam you have something
991 that is tiny, feels a lot like a unified kernel, but all it does is chainload
992 the real kernel. benefit: downloading these stubs would be tiny and quick,
993 hence cheap for enumeration.
995 * sysext: measure all activated sysext into a TPM PCR
997 * systemd-dissect: show available versions inside of a disk image, i.e. if
998 multiple versions are around of the same resource, show which ones. (in other
999 words: show partition labels).
1001 * maybe add a generator that reads /proc/cmdline, looks for
1002 systemd.pull-raw-portable=, systemd-pull-raw-sysext= and similar switches
1003 that take a URL as parameter. It then generates service units for
1004 systemd-pull calls that download these URLs if not installed yet. usecase:
1005 invoke a VM or nspawn container in a way it automatically deploys/runs these
1006 images as OS payloads. i.e. have a generic OS image you can point to any
1007 payload you like, which is then downloaded, securely verified and run.
1009 * improve scope units to support creation by pidfd instead of by PID
1011 * deprecate cgroupsv1 further (print log message at boot)
1013 * systemd-dissect: add --cat switch for dumping files such as /etc/os-release
1015 * per-service sandboxing option: ProtectIds=. If used, will overmount
1016 /etc/machine-id and /proc/sys/kernel/random/boot_id with synthetic files, to
1017 make it harder for the service to identify the host. Depending on the user
1018 setting it should be fully randomized at invocation time, or a hash of the
1019 real thing, keyed by the unit name or so. Of course, there are other ways to
1020 get these IDs (e.g. journal) or similar ids (e.g. MAC addresses, DMI ids, CPU
1021 ids), so this knob would only be useful in combination with other lockdown
1022 options. Particularly useful for portable services, and anything else that
1023 uses RootDirectory= or RootImage=. (Might also over-mount
1024 /sys/class/dmi/id/*{uuid,serial} with /dev/null).
1027 - add --all switch for rerunning kernel-install for all installed kernels
1029 * doc: prep a document explaining resolved's internal objects, i.e. Query
1030 vs. Question vs. Transaction vs. Stream and so on.
1032 * doc: prep a document explaining PID 1's internal logic, i.e. transactions,
1035 * bootspec: bring UEFI and userspace enumeration of bootspec entries back into
1036 sync, i.e. parse out architecture field in sd-boot (currently only done in
1039 * automatically ignore threaded cgroups in cg_xyz().
1041 * add linker script that implicitly adds symbol for build ID and new coredump
1042 json package metadata, and use that when logging
1044 * Enable RestrictFileSystems= for all our long-running services (similar:
1045 RestrictNetworkInterfaces=)
1047 * Add systemd-analyze security checks for RestrictFileSystems= and
1048 RestrictNetworkInterfaces=
1050 * cryptsetup/homed: implement TOTP authentication backed by TPM2 and its
1053 * man: rework os-release(5), and clearly separate our extension-release.d/ and
1054 initrd-release parts, i.e. list explicitly which fields are about what.
1056 * sysext: before applying a sysext, do a superficial validation run so that
1057 things are not rearranged to wildy. I.e. protect against accidental fuckups,
1058 such as masking out /usr/lib/ or so. We should probably refuse if existing
1059 inodes are replaced by other types of inodes or so.
1061 * userdb: when synthesizing NSS records, pick "best" password from defined
1062 passwords, not just the first. i.e. if there are multiple defined, prefer
1063 unlocked over locked and prefer non-empty over empty.
1065 * maybe add a tool inspired by the GPT auto discovery spec that runs in the
1066 initrd and rearranges the rootfs hierarchy via bind mounts, if
1067 enabled. Specifically in some top-level dir /@auto/ it will look for
1068 dirs/symlinks/subvolumes that are named after their purpose, and optionally
1069 encode a version as well as assessment counters, and then mount them into the
1070 file system tree to boot into, similar to how we do that for the gpt auto
1071 logic. Maybe then bind mount the original root into /.superior or something
1072 like that (so that update tools can look there). Further discussion in this
1074 https://lists.freedesktop.org/archives/systemd-devel/2021-November/047059.html
1075 The GPT dissection logic should automatically enable this tool whenever we
1076 detect a specially marked root fs (i.e introduce a new generic root gpt type
1077 for this, that is arch independent). The also implement this in the image
1078 dissection logic, so that nspawn/RootImage= and so on grok it. Maybe make
1079 generic enough so that it can also work for ostrees arrangements.
1081 * if a path ending in ".auto.d/" is set for RootDirectory=/RootImage= then do a
1082 strverscmp() of everything inside that dir and use that. i.e. implement very
1083 simple version control. Also use this in systemd-nspawn --image= and so on.
1085 * homed: while a home dir is not activated generate slightly different NSS
1086 records for it, that reports the home dir as "/" and the shell as some binary
1087 provided by us. Then, when an SSH login happens and SSH permits it our binary
1088 is invoked. This binary can then talk to homed and activate the homedir if
1089 it's not around yet, prompting the user for a password. Once that succeeded
1090 we'll switch to the real user record, i.e. home dir and shell, and our tool
1091 exec()s the latter. Net effect: ssh'ing into a homed account will just work:
1092 we'll neatly prompt for the homedir's password if its needed. –– Building on
1093 this we could take this even further: since this tool will potentially have
1094 access to the client's ssh-agent (if ssh-agent forwarding is enabled) we
1095 could implement SSH unlocking of a homedir with that: when enrolling a new
1096 ssh pubkey in a user record we'd ask the ssh-agent to sign some random value
1097 with the privkey, then use that as luks key to unlock the home dir. Will not
1098 work for ECDSA keys since their signatures contain a random component, but
1099 will work for RSA and Ed25519 keys.
1101 * add tiny service that decrypts encrypted user records passed via initrd
1102 credential logic and drops them into /run where nss-systemd can pick them up,
1103 similar to /run/host/userdb/. Usecase: drop a root user JSON record there,
1104 and use it in the initrd to log in as root with locally selected password,
1105 for debugging purposes. Other usecase: boot into qemu with regular user
1106 mounted from host. maybe put this in systemd-user-sessions.service?
1108 * drop dependency on libcap, replace by direct syscalls based on
1109 CapabilityQuintet we already have. (This likely allows us to drop libcap
1110 dep in the base OS image)
1112 * add concept for "exitrd" as inverse of "initrd", that we can transition to at
1113 shutdown, and has similar security semantics. This should then take the place
1114 of dracut's shutdown logic. Should probably support sysexts too. Care needs
1115 to be taken that the resulting logic ends up in RAM, i.e. is copied out of
1118 * userdbd: implement an additional varlink service socket that provides the
1119 host user db in restricted form, then allow this to be bind mounted into
1120 sandboxed environments that want the host database in minimal form. All
1121 records would be stripped of all meta info, except the basic UID/name
1122 info. Then use this in portabled environments that do not use PrivateUsers=1.
1124 * portabled: when extracting unit files and copying to system.attached, if a
1125 .p7s is available in the image, use it to protect the system.attached copy
1126 with fs-verity, so that it cannot be tampered with
1128 * logind introduce two types of sessions: "heavy" and "light". The former would
1129 be our current sessions. But the latter would be a new type of session that
1130 is mostly the same but does not pull in user@.service or wait for it. Then,
1131 allow configuration which type of session is desired via pam_systemd
1132 parameters, and then make user@.service's session one of these "light" ones.
1133 People could then choose to make FTP sessions and suchlike "light" if they
1134 don't want the service manager to be started for that.
1136 * /etc/veritytab: allow that the roothash column can be specified as fs path
1137 including a path to an AF_UNIX path, similar to how we do things with the
1138 keys of /etc/crypttab. That way people can store/provide the roothash
1139 externally and provide to us on demand only.
1141 * we probably should extend the root verity hash of the root fs into some PCR
1142 on boot. (i.e. maybe add a veritytab option tpm2-measure=12 or so to measure
1143 it into PCR 12); Similar: we probably should extend the LUKS volume key of
1144 the root fs into some PCR on boot. (i.e. maybe add a crypttab option
1145 tpm2-measure=15 or so to measure it into PCR 15); once both are in place
1146 update gpt-auto-discovery to generate these by default for the partitions it
1147 discovers. Static vendor stuff should probably end up in PCR 12 (i.e. the
1148 verity hash), with local keys in PCR 15 (i.e. the encryption volume
1149 key). That way, we nicely distinguish resources supplied by the OS vendor
1150 (i.e. sysext, root verity) from those inherently local (i.e. encryption key),
1151 which is useful if they shall be signed separately.
1153 * in uefi stub: query firmware regarding which PCR banks are being used, store
1154 that in EFI var. then use this when enrolling TPM2 in cryptsetup to verify
1155 that the selected PCRs actually are used by firmware.
1157 * rework recursive read-only remount to use new mount API
1159 * PAM: pick up authentication token from credentials
1161 * when mounting disk images: if IMAGE_ID/IMAGE_VERSION is set in os-release
1162 data in the image, make sure the image filename actually matches this, so
1163 that images cannot be misused.
1165 * New udev block device symlink names:
1166 /dev/disk/by-parttypelabel/<pttype>-<ptlabel>. Use case: if pt label is used
1167 as partition image version string, this is a safe way to reference a specific
1168 version of a specific partition type, in particular where related partitions
1169 are processed (e.g. verity + rootfs both named "LennartOS_0.7").
1172 - add fuzzing to the pattern parser
1173 - support casync as download mechanism
1174 - "systemd-sysupdate update --all" support, that iterates through all components
1175 defined on the host, plus all images installed into /var/lib/machines/,
1176 /var/lib/portable/ and so on.
1177 - figure out what to do about system extensions (i.e. they need to imply an
1178 update component, since otherwise system extenion' sysupdate.d/ files would
1179 override the host's update files.)
1180 - Allow invocation with a single transfer definition, i.e. with
1181 --definitions= pointing to a file rather than a dir.
1182 - add ability to disable implicit decompression of downloaded artifacts,
1183 i.e. a Compress=no option in the transfer definitions
1185 * in sd-id128: also parse UUIDs in RFC4122 URN syntax (i.e. chop off urn:uuid: prefix)
1187 * DynamicUser= + StateDirectory= → use uid mapping mounts, too, in order to
1188 make dirs appear under right UID.
1190 * systemd-sysext: optionally, run it in initrd already, before transitioning
1191 into host, to open up possibility for services shipped like that.
1193 * introduce /dev/disk/root/* symlinks that allow referencing partitions on the
1194 disk the rootfs is on in a reasonably secure way. (or maybe: add
1195 /dev/gpt-auto-{home,srv,boot,…} similar in style to /dev/gpt-auto-root as we
1198 * whenever we receive fds via SCM_RIGHTS make sure none got dropped due to the
1199 reception limit the kernel silently enforces.
1201 * Add service unit setting ConnectStream= which takes IP addresses and connects to them.
1203 * Similar, Load= which takes literal data in text or base64 format, and puts it
1204 into a memfd, and passes that. This enables some fun stuff, such as embedding
1205 bash scripts in unit files, by combining Load= with ExecStart=/bin/bash
1208 * add a ConnectSocket= setting to service unit files, that may reference a
1209 socket unit, and which will connect to the socket defined therein, and pass
1210 the resulting fd to the service program via socket activation proto.
1212 * Add a concept of ListenStream=anonymous to socket units: listen on a socket
1213 that is deleted in the fs. Usecase would be with ConnectSocket= above.
1215 * importd: support image signature verification with PKCS#7 + OpenBSD signify
1216 logic, as alternative to crummy gpg
1218 * add "systemd-analyze debug" + AttachDebugger= in unit files: The former
1219 specifies a command to execute; the latter specifies that an already running
1220 "systemd-analyze debug" instance shall be contacted and execution paused
1221 until it gives an OK. That way, tools like gdb or strace can be safely be
1222 invoked on processes forked off PID 1.
1224 * expose MS_NOSYMFOLLOW in various places
1226 * credentials system:
1227 - acquire from EFI variable?
1228 - acquire via ask-password?
1229 - acquire creds via keyring?
1230 - pass creds via keyring?
1231 - pass creds via memfd?
1232 - acquire + decrypt creds from pkcs11?
1233 - make systemd-cryptsetup acquire pw via creds logic
1234 - make PAMName= acquire pw via creds logic
1235 - make macsec/wireguard code in networkd read key via creds logic
1236 - make gatwayd/remote read key via creds logic
1237 - add sd_notify() command for flushing out creds not needed anymore
1238 - make user manager instances create and use a user-specific key (the one in
1239 /var/lib is root-only) and add --user switch to systemd-creds to use it
1241 * add tpm.target or so which is delayed until TPM2 device showed up in case
1242 firmware indicates there is one.
1244 * TPM2: auto-reenroll in cryptsetup, as fallback for hosed firmware upgrades
1247 * introduce a new group to own TPM devices
1249 * cryptsetup: add option for automatically removing empty password slot on boot
1251 * cryptsetup: optionally, when run during boot-up and password is never
1252 entered, and we are on battery power (or so), power off machine again
1254 * cryptsetup: when waiting for FIDO2/PKCS#11 token, tell plymouth that, and
1255 allow plymouth to abort the waiting and enter pw instead
1257 * make cryptsetup lower --iter-time
1259 * cryptsetup: allow encoding key directly in /etc/crypttab, maybe with a
1260 "base64:" prefix. Useful in particular for pkcs11 mode.
1262 * cryptsetup: reimplement the mkswap/mke2fs in cryptsetup-generator to use
1263 systemd-makefs.service instead.
1266 - cryptsetup-generator: allow specification of passwords in crypttab itself
1267 - support rd.luks.allow-discards= kernel cmdline params in cryptsetup generator
1269 * systemd-analyze netif that explains predictable interface (or networkctl)
1271 * Add service setting to run a service within the specified VRF. i.e. do the
1272 equivalent of "ip vrf exec".
1274 * special case some calls of chase() to use openat2() internally, so
1275 that the kernel does what we otherwise do.
1277 * add a new flag to chase() that stops chasing once the first missing
1278 component is found and then allows the caller to create the rest.
1280 * make use of new glibc 2.32 APIs sigabbrev_np() and strerrorname_np().
1282 * if /usr/bin/swapoff fails due to OOM, log a friendly explanatory message about it
1284 * pid1: Move to tracking of main pid/control pid of units per pidfd
1286 * pid1: support new clone3() fork-into-cgroup feature
1288 * pid1: also remove PID files of a service when the service starts, not just
1291 * make us use dynamically fewer deps for containers in general purpose distros:
1292 o turn into dlopen() deps:
1293 - kmod-libs (only when called from PID 1)
1294 - libblkid (only in RootImage= handling in PID 1, but not elsewhere)
1295 - libpam (only when called from PID 1)
1296 - bzip2, xz, lz4 (always — gzip and zstd should probably stay static deps the way they are,
1297 since they are so basic and our defaults)
1298 o move into separate libsystemd-shared-iptables.so .so
1299 - iptables-libs (only used by nspawn + networkd)
1301 * seccomp: maybe use seccomp_merge() to merge our filters per-arch if we can.
1302 Apparently kernel performance is much better with fewer larger seccomp
1303 filters than with more smaller seccomp filters.
1305 * systemd-path: add ESP and XBOOTLDR path. Add "private" runtime/state/cache dir enum,
1306 mapping to $RUNTIME_DIRECTORY, $STATE_DIRECTORY and such
1308 * seccomp: by default mask x32 ABI system wide on x86-64. it's on its way out
1310 * seccomp: don't install filters for ABIs that are masked anyway for the
1313 * busctl: maybe expose a verb "ping" for pinging a dbus service to see if it
1314 exists and responds.
1316 * socket units: allow creating a udev monitor socket with ListenDevices= or so,
1317 with matches, then activate app through that passing socket over
1320 - kill gnutls support in resolved
1321 - figure out what to do about libmicrohttpd, which has a hard dependency on
1323 - port fsprg over to a dlopen lib, then switch it to openssl
1325 * add growvol and makevol options for /etc/crypttab, similar to
1326 x-systemd.growfs and x-systemd-makefs.
1328 * userdb: allow username prefix searches in varlink API, allow realname and
1329 realname substr searches in varlink API
1331 * userdb: allow uid/gid range checks
1333 * userdb: allow existence checks
1335 * pid1: activation by journal search expression
1337 * when switching root from initrd to host, set the machine_id env var so that
1338 if the host has no machine ID set yet we continue to use the random one the
1341 * sd-event: add native support for P_ALL waitid() watching, then move PID 1 to
1342 it for reaping assigned but unknown children. This needs to some special care
1343 to operate somewhat sensibly in light of priorities: P_ALL will return
1344 arbitrary processes, regardless of the priority we want to watch them with,
1345 hence on each event loop iteration check all processes which we shall watch
1346 with higher prio explicitly, and then watch the entire rest with P_ALL.
1348 * tweak sd-event's child watching: keep a prioq of children to watch and use
1349 waitid() only on the children with the highest priority until one is waitable
1350 and ignore all lower-prio ones from that point on
1352 * maybe introduce xattrs that can be set on the root dir of the root fs
1353 partition that declare the volatility mode to use the image in. Previously I
1354 thought marking this via GPT partition flags but that's not ideal since
1355 that's outside of the LUKS encryption/verity verification, and we probably
1356 shouldn't operate in a volatile mode unless we got told so from a trusted
1359 * coredump: maybe when coredumping read a new xattr from /proc/$PID/exe that
1360 may be used to mark a whole binary as non-coredumpable. Would fix:
1361 https://bugs.freedesktop.org/show_bug.cgi?id=69447
1363 * teach parse_timestamp() timezones like the calendar spec already knows it
1365 * beef up s2h to implement a battery watch loop: instead of entering
1366 hibernation unconditionally after coming back from resume make a decision
1367 based on the battery load level: if battery level is above a specific
1368 threshold, go to suspend again, only hibernate if below it. This means we'd
1369 stick to suspend usually, but fall back to hibernation only when battery runs
1370 empty (well, subject to our sampling interval). Related to this, check if we
1371 can make ACPI _BTP (i.e. /sys/class/power_supply/*/alarm) work for us too,
1372 i.e. see if it can wake up machines from suspend, so that we could resume
1373 automatically when the system is low on power and move automatically to
1374 hibernation mode. (see
1375 https://uefi.org/sites/default/files/resources/ACPI%206_2_A_Sept29.pdf
1376 section 10.2.2.8 and
1377 https://docs.microsoft.com/en-us/windows-hardware/design/device-experiences/modern-standby-wake-sources
1380 * We should probably replace /etc/rc.d/README with a symlink to doc
1381 content. After all it is constant vendor data.
1383 * maybe add kernel cmdline params: to force random seed crediting
1385 * introduce a new per-process uuid, similar to the boot id, the machine id, the
1386 invocation id, that is derived from process creds, specifically a hashed
1387 combination of AT_RANDOM + getpid() + the starttime from
1388 /proc/self/status. Then add these ids implicitly when logging. Deriving this
1389 uuid from these three things has the benefit that it can be derived easily
1390 from /proc/$PID/ in a stable, and unique way that changes on both fork() and
1393 * let's not GC a unit while its ratelimits are still pending
1395 * when killing due to service watchdog timeout maybe detect whether target
1396 process is under ptracing and then log loudly and continue instead.
1398 * make rfkill uaccess controllable by default, i.e. steal rule from
1399 gnome-bluetooth and friends
1401 * make MAINPID= message reception checks even stricter: if service uses User=,
1402 then check sending UID and ignore message if it doesn't match the user or
1405 * maybe trigger a uevent "change" on a device if "systemctl reload xyz.device"
1408 * when importing an fs tree with machined, optionally apply userns-rec-chown
1410 * when importing an fs tree with machined, complain if image is not an OS
1412 * Maybe introduce a helper safe_exec() or so, which is to execve() which
1413 safe_fork() is to fork(). And then make revert the RLIMIT_NOFILE soft limit
1414 to 1K implicitly, unless explicitly opted-out.
1416 * rework seccomp/nnp logic that even if User= is used in combination with
1417 a seccomp option we don't have to set NNP. For that, change uid first whil
1418 keeping CAP_SYS_ADMIN, then apply seccomp, the drop cap.
1420 * when no locale is configured, default to UEFI's PlatformLang variable
1422 * add a new syscall group "@esoteric" for more esoteric stuff such as bpf() and
1423 usefaultd() and make systemd-analyze check for it.
1425 * paranoia: whenever we process passwords, call mlock() on the memory
1426 first. i.e. look for all places we use free_and_erasep() and
1427 augment them with mlock(). Also use MADV_DONTDUMP.
1428 Alternatively (preferably?) use memfd_secret().
1430 * Move RestrictAddressFamily= to the new cgroup create socket
1432 * optionally: turn on cgroup delegation for per-session scope units
1434 * sd-boot: optionally, show boot menu when previous default boot item has
1435 non-zero "tries done" count
1437 * augment CODE_FILE=, CODE_LINE= with something like CODE_BASE= or so which
1438 contains some identifier for the project, which allows us to include
1439 clickable links to source files generating these log messages. The identifier
1440 could be some abberviated URL prefix or so (taking inspiration from Go
1441 imports). For example, for systemd we could use
1442 CODE_BASE=github.com/systemd/systemd/blob/98b0b1123cc or so which is
1443 sufficient to build a link by prefixing "http://" and suffixing the
1446 * Augment MESSAGE_ID with MESSAGE_BASE, in a similar fashion so that we can
1447 make clickable links from log messages carrying a MESSAGE_ID, that lead to
1448 some explanatory text online.
1450 * maybe extend .path units to expose fanotify() per-mount change events
1452 * When reloading configuration PID 1 should reset all its properties to the
1453 original defaults before calling parse_config()
1455 * hibernate/s2h: if swap is on weird storage and refuse if so
1457 * cgroups: use inotify to get notified when somebody else modifies cgroups
1458 owned by us, then log a friendly warning.
1460 * beef up log.c with support for stripping ANSI sequences from strings, so that
1461 it is OK to include them in log strings. This would be particularly useful so
1462 that our log messages could contain clickable links for example for unit
1463 files and suchlike we operate on.
1465 * importd: add ability download images for portabled + sysext
1467 * add support for "portablectl attach http://foobar.com/waaa.raw (i.e. importd integration)
1469 * sync dynamic uids/gids between host+portable srvice (i.e. if DynamicUser=1 is set for a service, make sure that the
1470 selected user is resolvable in the service even if it ships its own /etc/passwd)
1472 * Fix DECIMAL_STR_MAX or DECIMAL_STR_WIDTH. One includes a trailing NUL, the
1473 other doesn't. What a disaster. Probably to exclude it.
1475 * Check that users of inotify's IN_DELETE_SELF flag are using it properly, as
1476 usually IN_ATTRIB is the right way to watch deleted files, as the former only
1477 fires when a file is actually removed from disk, i.e. the link count drops to
1478 zero and is not open anymore, while the latter happens when a file is
1479 unlinked from any dir.
1481 * port systemctl, busctl, … over to format-table.[ch]'s table formatters
1483 * pid1: lock image configured with RootDirectory=/RootImage= using the usual nspawn semantics while the unit is up
1485 * add --vacuum-xyz options to coredumpctl, matching those journalctl already has.
1487 * add CopyFile= or so as unit file setting that may be used to copy files or
1488 directory trees from the host to the services RootImage= and RootDirectory=
1489 environment. Which we can use for /etc/machine-id and in particular
1490 /etc/resolv.conf. Should be smart and do something useful on read-only
1491 images, for example fall back to read-only bind mounting the file instead.
1493 * show invocation ID in systemd-run output
1495 * bypass SIGTERM state in unit files if KillSignal is SIGKILL
1497 * add proper dbus APIs for the various sd_notify() commands, such as MAINPID=1
1498 and so on, which would mean we could report errors and such.
1500 * introduce DefaultSlice= or so in system.conf that allows changing where we
1501 place our units by default, i.e. change system.slice to something
1502 else. Similar, ManagerSlice= should exist so that PID1's own scope unit could
1503 be moved somewhere else too. Finally machined and logind should get similar
1504 options so that it is possible to move user session scopes and machines to a
1505 different slice too by default. Usecase: people who want to put resources on
1506 the entire system, with the exception of one specific service. See:
1507 https://lists.freedesktop.org/archives/systemd-devel/2018-February/040369.html
1509 * maybe rework get_user_creds() to query the user database if $SHELL is used
1510 for root, but only then.
1512 * calenderspec: add support for week numbers and day numbers within a
1513 year. This would allow us to define "bi-weekly" triggers safely.
1515 * sd-bus: add vtable flag, that may be used to request client creds implicitly
1516 and asynchronously before dispatching the operation
1518 * sd-bus: parse addresses given in sd_bus_set_addresses immediately and not
1519 only when used. Add unit tests.
1521 * make use of ethtool veth peer info in machined, for automatically finding out
1522 host-side interface pointing to the container.
1524 * add some special mode to LogsDirectory=/StateDirectory=… that allows
1525 declaring these directories without necessarily pulling in deps for them, or
1526 creating them when starting up. That way, we could declare that
1527 systemd-journald writes to /var/log/journal, which could be useful when we
1528 doing disk usage calculations and so on.
1530 * deprecate RootDirectoryStartOnly= in favour of a new ExecStart= prefix char
1532 * support projid-based quota in machinectl for containers
1534 * add a way to lock down cgroup migration: a boolean, which when set for a unit
1535 makes sure the processes in it can never migrate out of it
1537 * blog about fd store and restartable services
1539 * document Environment=SYSTEMD_LOG_LEVEL=debug drop-in in debugging document
1541 * rework ExecOutput and ExecInput enums so that EXEC_OUTPUT_NULL loses its
1542 magic meaning and is no longer upgraded to something else if set explicitly.
1544 * in the long run: permit a system with /etc/machine-id linked to /dev/null, to
1545 make it lose its identity, i.e. be anonymous. For this we'd have to patch
1546 through the whole tree to make all code deal with the case where no machine
1549 * optionally, collect cgroup resource data, and store it in per-unit RRD files,
1550 suitable for processing with rrdtool. Add bus API to access this data, and
1551 possibly implement a CPULoad property based on it.
1553 * beef up pam_systemd to take unit file settings such as cgroups properties as
1556 * maybe hook up xfs/ext4 quotactl() with services? i.e. automatically manage
1557 the quota of the user indicated in User= via unit file settings, like the
1558 other resource management concepts. Would mix nicely with DynamicUser=1. Or
1559 alternatively, do this with projids, so that we can also cover services
1560 running as root. Quota should probably cover all the special dirs such as
1561 StateDirectory=, LogsDirectory=, CacheDirectory=, as well as RootDirectory= if it
1562 is set, plus the whole disk space any image configured with RootImage=.
1564 * In DynamicUser= mode: before selecting a UID, use disk quota APIs on relevant
1565 disks to see if the UID is already in use.
1567 * expose IO accounting data on the bus, show it in systemd-run --wait and log
1568 about it in the resource log message
1570 * Add AddUser= setting to unit files, similar to DynamicUser=1 which however
1571 creates a static, persistent user rather than a dynamic, transient user. We
1572 can leverage code from sysusers.d for this.
1574 * add some optional flag to ReadWritePaths= and friends, that has the effect
1575 that we create the dir in question when the service is started. Example:
1577 ReadWritePaths=:/var/lib/foobar
1579 * Add ExecMonitor= setting. May be used multiple times. Forks off a process in
1580 the service cgroup, which is supposed to monitor the service, and when it
1581 exits the service is considered failed by its monitor.
1583 * track the per-service PAM process properly (i.e. as an additional control
1584 process), so that it may be queried on the bus and everything.
1586 * add a new "debug" job mode, that is propagated to unit_start() and for
1587 services results in two things: we raise SIGSTOP right before invoking
1588 execve() and turn off watchdog support. Then, use that to implement
1589 "systemd-gdb" for attaching to the start-up of any system service in its
1592 * gpt-auto logic: support encrypted swap, add kernel cmdline option to force
1593 it, and honour a gpt bit about it, plus maybe a configuration file
1595 * add a percentage syntax for TimeoutStopSec=, e.g. TimeoutStopSec=150%, and
1596 then use that for the setting used in user@.service. It should be understood
1597 relative to the configured default value.
1599 * enable LockMLOCK to take a percentage value relative to physical memory
1601 * Permit masking specific netlink APIs with RestrictAddressFamily=
1603 * define gpt header bits to select volatility mode
1605 * ProtectClock= (drops CAP_SYS_TIMES, adds seecomp filters for settimeofday, adjtimex), sets DeviceAllow o /dev/rtc
1607 * ProtectTracing= (drops CAP_SYS_PTRACE, blocks ptrace syscall, makes /sys/kernel/tracing go away)
1609 * ProtectMount= (drop mount/umount/pivot_root from seccomp, disallow fuse via DeviceAllow, imply Mountflags=slave)
1611 * ProtectKeyRing= to take keyring calls away
1613 * RemoveKeyRing= to remove all keyring entries of the specified user
1615 * ProtectReboot= that masks reboot() and kexec_load() syscalls, prohibits kill
1616 on PID 1 with the relevant signals, and makes relevant files in /sys and
1617 /proc (such as the sysrq stuff) unavailable
1619 * Support ReadWritePaths/ReadOnlyPaths/InaccessiblePaths in systemd --user instances
1620 via the new unprivileged Landlock LSM (https://landlock.io)
1622 * make sure the ratelimit object can deal with USEC_INFINITY as way to turn off things
1624 * in nss-systemd, if we run inside of RootDirectory= with PrivateUsers= set,
1625 find a way to map the User=/Group= of the service to the right name. This way
1626 a user/group for a service only has to exist on the host for the right
1629 * add bus API for creating unit files in /etc, reusing the code for transient units
1631 * add bus API to remove unit files from /etc
1633 * add bus API to retrieve current unit file contents (i.e. implement "systemctl cat" on the bus only)
1635 * rework fopen_temporary() to make use of open_tmpfile_linkable() (problem: the
1636 kernel doesn't support linkat() that replaces existing files, currently)
1638 * transient units: don't bother with actually setting unit properties, we
1639 reload the unit file anyway
1641 * optionally, also require WATCHDOG=1 notifications during service start-up and shutdown
1643 * cache sd_event_now() result from before the first iteration...
1645 * PID1: find a way how we can reload unit file configuration for
1646 specific units only, without reloading the whole of systemd
1648 * add an explicit parser for LimitRTPRIO= that verifies
1649 the specified range and generates sane error messages for incorrect
1652 * when we detect that there are waiting jobs but no running jobs, do something
1654 * PID 1 should send out sd_notify("WATCHDOG=1") messages (for usage in the --user mode, and when run via nspawn)
1656 * there's probably something wrong with having user mounts below /sys,
1657 as we have for debugfs. for example, src/core/mount.c handles mounts
1658 prefixed with /sys generally special.
1659 https://lists.freedesktop.org/archives/systemd-devel/2015-June/032962.html
1661 * fstab-generator: default to tmpfs-as-root if only usr= is specified on the kernel cmdline
1663 * docs: bring https://www.freedesktop.org/wiki/Software/systemd/MyServiceCantGetRealtime up to date
1665 * add a job mode that will fail if a transaction would mean stopping
1666 running units. Use this in timedated to manage the NTP service
1668 https://lists.freedesktop.org/archives/systemd-devel/2015-April/030229.html
1670 * The udev blkid built-in should expose a property that reflects
1671 whether media was sensed in USB CF/SD card readers. This should then
1672 be used to control SYSTEMD_READY=1/0 so that USB card readers aren't
1673 picked up by systemd unless they contain a medium. This would mirror
1674 the behaviour we already have for CD drives.
1676 * hostnamectl: show root image uuid
1678 * Find a solution for SMACK capabilities stuff:
1679 https://lists.freedesktop.org/archives/systemd-devel/2014-December/026188.html
1681 * synchronize console access with BSD locks:
1682 https://lists.freedesktop.org/archives/systemd-devel/2014-October/024582.html
1684 * as soon as we have sender timestamps, revisit coalescing multiple parallel daemon reloads:
1685 https://lists.freedesktop.org/archives/systemd-devel/2014-December/025862.html
1687 * figure out when we can use the coarse timers
1689 * maybe allow timer units with an empty Units= setting, so that they
1690 can be used for resuming the system but nothing else.
1692 * what to do about udev db binary stability for apps? (raw access is not an option)
1694 * exponential backoff in timesyncd when we cannot reach a server
1696 * timesyncd: add ugly bus calls to set NTP servers per-interface, for usage by NM
1698 * add systemd.abort_on_kill or some other such flag to send SIGABRT instead of SIGKILL
1699 (throughout the codebase, not only PID1)
1701 * drop nss-myhostname in favour of nss-resolve?
1705 - service registration
1706 - service/domain/types browsing
1708 - DNS-SD service registration from socket units
1709 - resolved should optionally register additional per-interface LLMNR
1710 names, so that for the container case we can establish the same name
1711 (maybe "host") for referencing the server, everywhere.
1712 - allow clients to request DNSSEC for a single lookup even if DNSSEC is off (?)
1713 - hook up resolved with machined-based address resolution
1715 * refcounting in sd-resolve is borked
1717 * add new gpt type for btrfs volumes
1719 * generator that automatically discovers btrfs subvolumes, identifies their purpose based on some xattr on them.
1721 * a way for container managers to turn off getty starting via $container_headless= or so...
1723 * figure out a nice way how we can let the admin know what child/sibling unit causes cgroup membership for a specific unit
1725 * For timer units: add some mechanisms so that timer units that trigger immediately on boot do not have the services
1726 they run added to the initial transaction and thus confuse Type=idle.
1728 * add bus api to query unit file's X fields.
1730 * gpt-auto-generator:
1731 - Define new partition type for encrypted swap? Support probed LUKS for encrypted swap?
1732 - Make /home automount rather than mount?
1734 * add generator that pulls in systemd-network from containers when
1735 CAP_NET_ADMIN is set, more than the loopback device is defined, even
1736 when it is otherwise off
1738 * MessageQueueMessageSize= (and suchlike) should use parse_iec_size().
1740 * implement Distribute= in socket units to allow running multiple
1741 service instances processing the listening socket, and open this up
1745 - implement per-slice CPUFairScheduling=1 switch
1746 - introduce high-level settings for RT budget, swappiness
1747 - how to reset dynamically changed unit cgroup attributes sanely?
1748 - when reloading configuration, apply new cgroup configuration
1749 - when recursively showing the cgroup hierarchy, optionally also show
1750 the hierarchies of child processes
1751 - add settings for cgroup.max.descendants and cgroup.max.depth,
1752 maybe use them for user@.service
1755 - add field to transient units that indicate whether systemd or somebody else saves/restores its settings, for integration with libvirt
1757 * when we detect low battery and no AC on boot, show pretty splash and refuse boot
1759 * libsystemd-journal, libsystemd-login, libudev: add calls to easily attach these objects to sd-event event loops
1761 * be more careful what we export on the bus as (usec_t) 0 and (usec_t) -1
1763 * rfkill,backlight: we probably should run the load tools inside of the udev rules so that the state is properly initialized by the time other software sees it
1765 * After coming back from hibernation reset hibernation swap partition using the /dev/snapshot ioctl APIs
1767 * If we try to find a unit via a dangling symlink, generate a clean
1768 error. Currently, we just ignore it and read the unit from the search
1771 * refuse boot if /usr/lib/os-release is missing or /etc/machine-id cannot be set up
1773 * man: the documentation of Restart= currently is very misleading and suggests the tools from ExecStartPre= might get restarted.
1775 * load .d/*.conf dropins for device units
1777 * There's currently no way to cancel fsck (used to be possible via C-c or c on the console)
1779 * add option to sockets to avoid activation. Instead just drop packets/connections, see http://cyberelk.net/tim/2012/02/15/portreserve-systemd-solution/
1781 * make sure systemd-ask-password-wall does not shutdown systemd-ask-password-console too early
1783 * verify that the AF_UNIX sockets of a service in the fs still exist
1784 when we start a service in order to avoid confusion when a user
1785 assumes starting a service is enough to make it accessible
1787 * Make it possible to set the keymap independently from the font on
1788 the kernel cmdline. Right now setting one resets also the other.
1790 * and a dbus call to generate target from current state
1792 * investigate whether the gnome pty helper should be moved into systemd, to provide cgroup support.
1794 * dot output for --test showing the 'initial transaction'
1796 * be able to specify a forced restart of service A where service B depends on, in case B
1797 needs to be auto-respawned?
1800 - When logging about multiple units (stopping BoundTo units, conflicts, etc.),
1801 log both units as UNIT=, so that journalctl -u triggers on both.
1802 - generate better errors when people try to set transient properties
1803 that are not supported...
1804 https://lists.freedesktop.org/archives/systemd-devel/2015-February/028076.html
1805 - maybe introduce WantsMountsFor=? Usecase:
1806 https://lists.freedesktop.org/archives/systemd-devel/2015-January/027729.html
1807 - recreate systemd's D-Bus private socket file on SIGUSR2
1808 - move PAM code into its own binary
1809 - when we automatically restart a service, ensure we restart its rdeps, too.
1810 - hide PAM options in fragment parser when compile time disabled
1811 - Support --test based on current system state
1812 - If we show an error about a unit (such as not showing up) and it has no Description string, then show a description string generated form the reverse of unit_name_mangle().
1813 - after deserializing sockets in socket.c we should reapply sockopts and things
1814 - drop PID 1 reloading, only do reexecing (difficult: Reload()
1815 currently is properly synchronous, Reexec() is weird, because we
1816 cannot delay the response properly until we are back, so instead of
1817 being properly synchronous we just keep open the fd and close it
1818 when done. That means clients do not get a successful method reply,
1819 but much rather a disconnect on success.
1820 - when breaking cycles drop sysv services first, then services from /run, then from /etc, then from /usr
1821 - when a bus name of a service disappears from the bus make sure to queue further activation requests
1822 - maybe introduce CoreScheduling=yes/no to optionally set a PR_SCHED_CORE cookie, so that all
1823 processes in a service's cgroup share the same cookie and are guaranteed not to share SMT cores
1824 with other units https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/admin-guide/hw-vuln/core-scheduling.rst
1827 - allow port=0 in .socket units
1828 - maybe introduce ExecRestartPre=
1829 - implement Register= switch in .socket units to enable registration
1830 in Avahi, RPC and other socket registration services.
1831 - allow Type=simple with PIDFile=
1832 https://bugzilla.redhat.com/show_bug.cgi?id=723942
1833 - allow writing multiple conditions in unit files on one line
1834 - introduce Type=pid-file
1835 - add a concept of RemainAfterExit= to scope units
1836 - Allow multiple ExecStart= for all Type= settings, so that we can cover rescue.service nicely
1837 - add verification of [Install] section to systemd-analyze verify
1840 - timer units should get the ability to trigger when DST changes
1841 - Modulate timer frequency based on battery state
1843 * add libsystemd-password or so to query passwords during boot using the password agent logic
1845 * clean up date formatting and parsing so that all absolute/relative timestamps we format can also be parsed
1847 * on shutdown: move utmp, wall, audit logic all into PID 1 (or logind?), get rid of systemd-update-utmp-runlevel
1849 * make repeated alt-ctrl-del presses printing a dump
1851 * currently x-systemd.timeout is lost in the initrd, since crypttab is copied into dracut, but fstab is not
1853 * add a pam module that passes the hdd passphrase into the PAM stack and then expires it, for usage by gdm auto-login.
1855 * add a pam module that on password changes updates any LUKS slot where the password matches
1858 - add unit tests for config_parse_device_allow()
1860 * seems that when we follow symlinks to units we prefer the symlink
1861 destination path over /etc and /usr. We should not do that. Instead
1862 /etc should always override /run+/usr and also any symlink
1865 * when isolating, try to figure out a way how we implicitly can order
1866 all units we stop before the isolating unit...
1868 * teach ConditionKernelCommandLine= globs or regexes (in order to match foobar={no,0,off})
1870 * Add ConditionDirectoryNotEmpty= handle non-absoute paths as a search path or add
1871 ConditionConfigSearchPathNotEmpty= or different syntax? See the discussion starting at
1872 https://github.com/systemd/systemd/pull/15109#issuecomment-607740136.
1874 * BootLoaderSpec: Define a way how an installer can figure out whether a BLS
1875 compliant boot loader is installed.
1877 * think about requeuing jobs when daemon-reload is issued? usecase:
1878 the initrd issues a reload after fstab from the host is accessible
1879 and we might want to requeue the mounts local-fs acquired through
1882 * systemd-inhibit: make taking delay locks useful: support sending SIGINT or SIGTERM on PrepareForSleep()
1884 * remove any syslog support from log.c — we probably cannot do this before split-off udev is gone for good
1886 * shutdown logging: store to EFI var, and store to USB stick?
1888 * merge unit_kill_common() and unit_kill_context()
1890 * add a dependency on standard-conf.xml and other included files to man pages
1892 * MountFlags=shared acts as MountFlags=slave right now.
1894 * properly handle loop back mounts via fstab, especially regards to fsck/passno
1896 * initialize the hostname from the fs label of /, if /etc/hostname does not exist?
1900 - GetAllProperties() on a non-existing object does not result in a failure currently
1901 - port to sd-resolve for connecting to TCP dbus servers
1902 - 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
1903 - see if we can drop more message validation on the sending side
1904 - add API to clone sd_bus_message objects
1905 - longer term: priority inheritance
1906 - dbus spec updates:
1907 - NameLost/NameAcquired obsolete
1909 - update systemd.special(7) to mention that dbus.socket is only about the compatibility socket now
1912 - allow multiple signal handlers per signal?
1913 - document chaining of signal handler for SIGCHLD and child handlers
1914 - define more intervals where we will shift wakeup intervals around in, 1h, 6h, 24h, ...
1915 - maybe support iouring as backend, so that we allow hooking read and write
1916 operations instead of IO ready events into event loops. See considerations
1918 http://blog.vmsplice.net/2020/07/rethinking-event-loop-integration-for.html
1920 * dbus: when a unit failed to load (i.e. is in UNIT_ERROR state), we
1921 should be able to safely try another attempt when the bus call LoadUnit() is invoked.
1923 * document org.freedesktop.MemoryAllocation1
1925 * maybe do not install getty@tty1.service symlink in /etc but in /usr?
1927 * print a nicer explanation if people use variable/specifier expansion in ExecStart= for the first word
1929 * mount: turn dependency information from /proc/self/mountinfo into dependency information between systemd units.
1932 - honor language efi variables for default language selection (if there are any?)
1933 - honor timezone efi variables for default timezone selection (if there are any?)
1934 - change bootctl to be backed by systemd-bootd to control temporary and persistent default boot goal plus efi variables
1936 - recognize the case when not booted on EFI
1938 * bootctl,sd-boot: actually honour the "architecture" key
1941 - show whether UEFI audit mode is available
1942 - teach it to prepare an ESP wholesale, i.e. with mkfs.vfat invocation
1943 - teach it to copy in unified kernel images and maybe type #1 boot loader spec entries from host
1946 - logind: optionally, ignore idle-hint logic for autosuspend, block suspend as long as a session is around
1947 - logind: wakelock/opportunistic suspend support
1948 - Add pretty name for seats in logind
1949 - logind: allow showing logout dialog from system?
1950 - add Suspend() bus calls which take timestamps to fix double suspend issues when somebody hits suspend and closes laptop quickly.
1951 - if pam_systemd is invoked by su from a process that is outside of a
1952 any session we should probably just become a NOP, since that's
1953 usually not a real user session but just some system code that just
1955 - logind: make the Suspend()/Hibernate() bus calls wait for the for
1956 the job to be completed. before returning, so that clients can wait
1957 for "systemctl suspend" to finish to know when the suspending is
1959 - logind: when the power button is pressed short, just popup a
1960 logout dialog. If it is pressed for 1s, do the usual
1961 shutdown. Inspiration are Macs here.
1962 - expose "Locked" property on logind session objects
1963 - maybe allow configuration of the StopTimeout for session scopes
1964 - rename session scope so that it includes the UID. THat way
1965 the session scope can be arranged freely in slices and we don't have
1966 make assumptions about their slice anymore.
1967 - follow PropertiesChanged state more closely, to deal with quick logouts and
1969 - (optionally?) spawn seat-manager@$SEAT.service whenever a seat shows up that as CanGraphical set
1970 - expose details of boot entries on the bus. In particular, it should be possible
1971 to query the list of boot entry titles that bootctl / sd-boot would show.
1972 Currently we only expose their identifiers.
1974 * move multiseat vid/pid matches from logind udev rule to hwdb
1976 * logind: rework pam_logind to also do a bus call in case of invocation from
1977 user@.service, which returns the XDG_RUNTIME_DIR value, and make this
1978 behaviour selectable via pam module option.
1980 * delay activation of logind until somebody logs in, or when /dev/tty0 pulls it
1981 in or lingering is on (so that containers don't bother with it until PAM is used). also exit-on-idle
1984 - consider introducing implicit _TTY= + _PPID= + _EUID= + _EGID= + _FSUID= + _FSGID= fields
1985 - journald: also get thread ID from client, plus thread name
1986 - journal: when waiting for journal additions in the client always sleep at least 1s or so, in order to minimize wakeups
1987 - add API to close/reopen/get fd for journal client fd in libsystemd-journal.
1988 - fall back to /dev/log based logging in libsystemd-journal, if we cannot log natively?
1989 - declare the local journal protocol stable in the wiki interface chart
1990 - sd-journal: speed up sd_journal_get_data() with transparent hash table in bg
1991 - journald: when dropping msgs due to ratelimit make sure to write
1992 "dropped %u messages" not only when we are about to print the next
1993 message that works, but already after a short timeout
1994 - check if we can make journalctl by default use --follow mode inside of less if called without args?
1995 - maybe add API to send pairs of iovecs via sd_journal_send
1996 - journal: add a setgid "systemd-journal" utility to invoke from libsystemd-journal, which passes fds via STDOUT and does PK access
1997 - journalctl: support negative filtering, i.e. FOOBAR!="waldo",
1998 and !FOOBAR for events without FOOBAR.
1999 - journal: store timestamp of journal_file_set_offline() in the header,
2000 so it is possible to display when the file was last synced.
2001 - journal-send.c, log.c: when the log socket is clogged, and we drop, count this and write a message about this when it gets unclogged again.
2002 - journal: find a way to allow dropping history early, based on priority, other rules
2003 - journal: When used on NFS, check payload hashes
2004 - journald: add kernel cmdline option to disable ratelimiting for debug purposes
2005 - refuse taking lower-case variable names in sd_journal_send() and friends.
2006 - journald: we currently rotate only after MaxUse+MaxFilesize has been reached.
2007 - journal: deal nicely with byte-by-byte copied files, especially regards header
2008 - journal: sanely deal with entries which are larger than the individual file size, but where the components would fit
2009 - Replace utmp, wtmp, btmp, and lastlog completely with journal
2010 - journalctl: instead --after-cursor= maybe have a --cursor=XYZ+1 syntax?
2011 - when a kernel driver logs in a tight loop, we should ratelimit that too.
2012 - journald: optionally, log debug messages to /run but everything else to /var
2013 - journald: when we drop syslog messages because the syslog socket is
2014 full, make sure to write how many messages are lost as first thing
2015 to syslog when it works again.
2016 - journald: allow per-priority and per-service retention times when rotating/vacuuming
2017 - journald: make use of uid-range.h to managed uid ranges to split
2019 - journalctl: add the ability to look for the most recent process of a binary. journalctl /usr/bin/X11 --pid=-1 or so...
2020 - improve journalctl performance by loading journal files
2021 lazily. Encode just enough information in the file name, so that we
2022 do not have to open it to know that it is not interesting for us, for
2023 the most common operations.
2024 - man: document that corrupted journal files is nothing to act on
2025 - rework journald sigbus stuff to use mutex
2026 - Set RLIMIT_NPROC for systemd-journal-xyz, and all other of our
2027 services that run under their own user ids, and use User= (but only
2028 in a world where userns is ubiquitous since otherwise we cannot
2029 invoke those daemons on the host AND in a container anymore). Also,
2030 if LimitNPROC= is used without User= we should warn and refuse
2032 - journalctl --verify: don't show files that are currently being
2033 written to as FAIL, but instead show that their are being written to.
2034 - add journalctl -H that talks via ssh to a remote peer and passes through
2036 - add a version of --merge which also merges /var/log/journal/remote
2037 - journalctl: -m should access container journals directly by enumerating
2038 them via machined, and also watch containers coming and going.
2039 Benefit: nspawn --ephemeral would start working nicely with the journal.
2040 - assign MESSAGE_ID to log messages about failed services
2041 - check if loop in decompress_blob_xz() is necessary
2043 * journald: support RFC3164 fully for the incoming syslog transport, see
2044 https://github.com/systemd/systemd/issues/19251#issuecomment-816601955
2046 * Hook up journald's FSS logic with TPM2: seal the verification disk by
2047 time-based policy, so that the verification key can remain on host and ve
2050 * rework journalctl -M to be based on a machined method that generates a mount
2051 fd of the relevant journal dirs in the container with uidmapping applied to
2052 allow the host to read it, while making everything read-only.
2054 * journald: add varlink service that allows subscribing to certain log events,
2055 for example matching by message ID, or log level returns a list of journal
2056 cursors as they happen.
2058 * journald: also collect CLOCK_BOOTTIME timestamps per log entry. Then, derive
2059 "corrected" CLOCK_REALTIME information on display from that and the timestamp
2060 info of the newest entry of the specific boot (as identified by the boot
2061 ID). This way, if a system comes up without a valid clock but acquires a
2062 better clock later, we can "fix" older entry timestamps on display, by
2063 calculating backwards. We cannot use CLOCK_MONOTONIC for this, since it does
2064 not account for suspend phases. This would then also enable us to correct the
2065 kmsg timestamping we consume (where we erroneously assume the clock was in
2066 CLOCK_MONOTONIC, but it actually is CLOCK_BOOTTIME as per kernel).
2068 * in journald, write out a recognizable log record whenever the system clock is
2069 changed ("stepped"), and in timesyncd whenever we acquire an NTP fix
2070 ("slewing"). Then, in journalctl for each boot time we come across, find
2071 these records, and use the structured info they include to display
2072 "corrected" wallclock time, as calculated from the monotonic timestamp in the
2073 log record, adjusted by the delta declared in the structured log record.
2075 * in journald: whenever we start a new journal file because the boot ID
2076 changed, let's generate a recognizable log record containing info about old
2077 and new ID. Then, when displaying log stream in journalctl look for these
2078 records, to be able to order them.
2080 * journald: generate recognizable log events whenever we shutdown journald
2081 cleanly, and when we migrate run → var. This way tools can verify that a
2082 previous boot terminated cleanly, because either of these two messages must
2083 be safely written to disk, then.
2085 * hook up journald with TPMs? measure new journal records to the TPM in regular
2086 intervals, validate the journal against current TPM state with that. (taking
2087 inspiration from IMA log)
2089 * sd-journal puts a limit on parallel journal files to view at once. journald
2090 should probably honour that same limit (JOURNAL_FILES_MAX) when vacuuming to
2091 ensure we never generate more files than we can actually view.
2093 * maybe add a tool that displays most recent journal logs as QR code to scan
2094 off screen and run it automatically on boot failures, emergency logs and
2095 such. Use DRM APIs directly, see
2096 https://github.com/dvdhrm/docs/blob/master/drm-howto/modeset.c for an example
2099 * maybe implicitly attach monotonic+realtime timestamps to outgoing messages in
2100 log.c and sd-journal-send
2102 * journalctl/timesyncd: whenever timesyncd acquires a synchronization from NTP,
2103 create a structured log entry that contains boot ID, monotonic clock and
2104 realtime clock (I mean, this requires no special work, as these three fields
2105 are implicit). Then in journalctl when attempting to display the realtime
2106 timestamp of a log entry, first search for the closest later log entry
2107 of this kinda that has a matching boot id, and convert the monotonic clock
2108 timestamp of the entry to the realtime clock using this info. This way we can
2109 retroactively correct the wallclock timestamps, in particular for systems
2110 without RTC, i.e. where initially wallclock timestamps carry rubbish, until
2111 an NTP sync is acquired.
2113 * introduce per-unit (i.e. per-slice, per-service) journal log size limits.
2115 * journald: do journal file writing out-of-process, with one writer process per
2116 client UID, so that synthetic hash table collisions can slow down a specific
2117 user's journal stream down but not the others.
2119 * tweak journald context caching. In addition to caching per-process attributes
2120 keyed by PID, cache per-cgroup attributes (i.e. the various xattrs we read)
2121 keyed by cgroup path, and guarded by ctime changes. This should provide us
2122 with a nice speed-up on services that have many processes running in the same
2125 * maybe add call sd_journal_set_block_timeout() or so to set SO_SNDTIMEO for
2126 the sd-journal logging socket, and, if the timeout is set to 0, sets
2127 O_NONBLOCK on it. That way people can control if and when to block for
2130 * journalctl: make sure -f ends when the container indicated by -M terminates
2132 * journald: sigbus API via a signal-handler safe function that people may call
2133 from the SIGBUS handler
2135 * add a test if all entries in the catalog are properly formatted.
2136 (Adding dashes in a catalog entry currently results in the catalog entry
2137 being silently skipped. journalctl --update-catalog must warn about this,
2138 and we should also have a unit test to check that all our message are OK.)
2140 * build short web pages out of each catalog entry, build them along with man
2141 pages, and include hyperlinks to them in the journal output
2144 - when user tries to log into record signed by unrecognized key, automatically add key to our chain after polkit auth
2145 - rollback when resize fails mid-operation
2146 - GNOME's side for forget key on suspend (requires rework so that lock screen runs outside of uid)
2147 - update LUKS password on login if we find there's a password that unlocks the JSON record but not the LUKS device.
2148 - create on activate?
2149 - properties: icon url?, preferred session type?, administrator bool (which translates to 'wheel' membership)?, address?, telephone?, vcard?, samba stuff?, parental controls?
2150 - communicate clearly when usb stick is safe to remove. probably involves
2151 beefing up logind to make pam session close hook synchronous and wait until
2152 systemd --user is shut down.
2153 - logind: maybe keep a "busy fd" as long as there's a non-released session around or the user@.service
2154 - maybe make automatic, read-only, time-based reflink-copies of LUKS disk
2155 images (and btrfs snapshots of subvolumes) (think: time machine)
2156 - distinguish destroy / remove (i.e. currently we can unregister a user, unregister+remove their home directory, but not just remove their home directory)
2157 - in systemd's PAMName= logic: query passwords with ssh-askpassword, so that we can make "loginctl set-linger" mode work
2158 - fingerprint authentication, pattern authentication, …
2159 - make sure "classic" user records can also be managed by homed
2160 - make size of $XDG_RUNTIME_DIR configurable in user record
2161 - query password from kernel keyring first
2162 - update even if record is "absent"
2163 - move acct mgmt stuff from pam_systemd_home to pam_systemd?
2164 - when "homectl --pkcs11-token-uri=" is used, synthesize ssh-authorized-keys records for all keys we have private keys on the stick for
2165 - make slice for users configurable (requires logind rework)
2166 - logind: populate auto-login list bus property from PKCS#11 token
2167 - when determining state of a LUKS home directory, check DM suspended sysfs file
2168 - when homed is in use, maybe start the user session manager in a mount namespace with MS_SLAVE,
2169 so that mounts propagate down but not up - eg, user A setting up a backup volume
2170 doesn't mean user B sees it
2171 - use credentials logic/TPM2 logic to store homed signing key
2172 - permit multiple user record signing keys to be used locally, and pick
2173 the right one for signing records automatically depending on a pre-existing
2175 - add a way to "adopt" a home directory, i.e. strip foreign signatures
2176 and insert a local signature instead.
2177 - as an extension to the directory+subvolume backend: if located on
2178 especially marked fs, then sync down password into LUKS header of that fs,
2179 and always verify passwords against it too. Bootstrapping is a problem
2180 though: if no one is logged in (or no other user even exists yet), how do you
2181 unlock the volume in order to create the first user and add the first pw.
2182 - support new FS_IOC_ADD_ENCRYPTION_KEY ioctl for setting up fscrypt
2183 - maybe pre-create ~/.cache as subvol so that it can have separate quota
2185 - add a switch to homectl (maybe called --first-boot) where it will check if
2186 any non-system users exist, and if not prompts interactively for basic user
2187 info, mimicking systemd-firstboot. Then, place this in a service that runs
2188 after systemd-homed, but before gdm and friends, as a simple, barebones
2189 fallback logic to get a regular user created on uninitialized systems.
2190 - store PKCS#11 + FIDO2 token info in LUKS2 header, compatible with
2191 systemd-cryptsetup, so that it can unlock homed volumes
2192 - maybe make all *.home files owned by `systemd-home` user or so, so that we
2193 can easily set overall quota for all users
2194 - on login, if we can't fallocate initially, but rebalance is on, then allow
2195 login in discard mode, then immediately rebalance, then turn off discard
2196 - extend user records with optional "bulk" data. Specifically, a user
2197 avatar/photo or so. This data should be stored along with the user record,
2198 but probably shouldn't be part of the record itself, since it might be
2201 * add a new switch --auto-definitions=yes/no or so to systemd-repart. If
2202 specified, synthesize a definition automatically if we can: enlarge last
2203 partition on disk, but only if it is marked for growing and not read-only.
2205 * systemd-repart: read LUKS encryption key from $CREDENTIALS_DIRECTORY
2207 * systemd-repart: add a switch to factory reset the partition table without
2208 immediately applying the new configuration again. i.e. --factory-reset=leave
2209 or so. (this is useful to factory reset an image, then putting it into
2210 another machine, ensuring that luks key is generated on new machine, not old)
2212 * systemd-repart: support setting up dm-integrity with HMAC
2214 * systemd-repart: maybe remove half-initialized image on failure. It fails
2215 if the output file exists, so a repeated invocation will usually fail if
2216 something goes wrong on the way.
2218 * systemd-repart: drop pager mode on normal operation?
2220 * systemd-repart: by default generate minimized partition tables (i.e. tables
2221 that only cover the space actually used, excluding any free space at the
2222 end), in order to maximize dd'ability. Requires libfdisk work, see
2223 https://github.com/karelzak/util-linux/issues/907
2225 * systemd-repart: MBR partition table support. Care needs to be taken regarding
2226 Type=, so that partition definitions can sanely apply to both the GPT and the
2227 MBR case. Idea: accept syntax "Type=gpt:home mbr:0x83" for setting the types
2228 for the two partition types explicitly. And provide an internal mapping so
2229 that "Type=linux-generic" maps to the right types for both partition tables
2232 * systemd-repart: allow sizing partitions as factor of available RAM, so that
2233 we can reasonably size swap partitions for hibernation.
2235 * systemd-repart: allow boolean option that ensures that if existing partition
2236 doesn't exist within the configured size bounds the whole command fails. This
2237 is useful to implement ESP vs. XBOOTLDR schemes in installers: have one set
2238 of repart files for the case where ESP is large enough and one where it isn't
2239 and XBOOTLDR is added in instead. Then apply the former first, and if it
2240 fails to apply use the latter.
2242 * systemd-repart: add per-partition option to never reuse existing partition
2243 and always create anew even if matching partition already exists.
2245 * systemd-repart: add per-partition option to fail if partition already exist,
2246 i.e. is not added new. Similar, add option to fail if partition does not exist yet.
2248 * systemd-repart: allow disabling growing of specific partitions, or making
2249 them (think ESP: we don't ever want to grow it, since we cannot resize vfat)
2250 Also add option to disable operation via kernel command line.
2252 * systemd-repart: make it a static checker during early boot for existence and
2253 absence of other partitions for trusted boot environments
2255 * systemd-repart: add support for SD_GPT_FLAG_GROWFS also on real systems, i.e.
2256 generate some unit to actually enlarge the fs after growing the partition
2259 * systemd-repart: do not print "Successfully resized …" when no change was done.
2262 - document that deps in [Unit] sections ignore Alias= fields in
2263 [Install] units of other units, unless those units are disabled
2264 - man: clarify that time-sync.target is not only sysv compat but also useful otherwise. Same for similar targets
2265 - document that service reload may be implemented as service reexec
2266 - add a man page containing packaging guidelines and recommending usage of things like Documentation=, PrivateTmp=, PrivateNetwork= and ReadOnlyDirectories=/etc /usr.
2267 - document systemd-journal-flush.service properly
2268 - documentation: recommend to connect the timer units of a service to the service via Also= in [Install]
2269 - man: document the very specific env the shutdown drop-in tools live in
2270 - man: add more examples to man pages,
2271 - in particular an example how to do the equivalent of switching runlevels
2272 - man: maybe sort directives in man pages, and take sections from --help and apply them to man too
2273 - document root=gpt-auto properly
2276 - add systemctl switch to dump transaction without executing it
2277 - Add a verbose mode to "systemctl start" and friends that explains what is being done or not done
2278 - "systemctl disable" on a static unit prints no message and does
2279 nothing. "systemctl enable" does nothing, and gives a bad message
2280 about it. Should fix both to print nice actionable messages.
2281 - print nice message from systemctl --failed if there are no entries shown, and hook that into ExecStartPre of rescue.service/emergency.service
2282 - add new command to systemctl: "systemctl system-reexec" which reexecs as many daemons as virtually possible
2283 - systemctl enable: fail if target to alias into does not exist? maybe show how many units are enabled afterwards?
2284 - systemctl: "Journal has been rotated since unit was started." message is misleading
2285 - systemctl status output should include list of triggering units and their status
2287 * introduce an option (or replacement) for "systemctl show" that outputs all
2288 properties as JSON, similar to busctl's new JSON output. In contrast to that
2289 it should skip the variant type string though.
2291 * Add a "systemctl list-units --by-slice" mode or so, which rearranges the
2292 output of "systemctl list-units" slightly by showing the tree structure of
2293 the slices, and the units attached to them.
2295 * add "systemctl wait" or so, which does what "systemd-run --wait" does, but
2296 for all units. It should be both a way to pin units into memory as well as a
2297 wait to retrieve their exit data.
2299 * show whether a service has out-of-date configuration in "systemctl status" by
2300 using mtime data of ConfigurationDirectory=.
2302 * "systemctl preset-all" should probably order the unit files it
2303 operates on lexicographically before starting to work, in order to
2304 ensure deterministic behaviour if two unit files conflict (like DMs
2307 * add "systemctl start -v foobar.service" that shows logs of a service
2308 while the start command runs. This is non-trivial to do without
2309 races though, since we should flush out all journal messages before
2310 returning from the "systemctl stop".
2312 * systemctl: if some operation fails, show log output?
2314 * Add a new verb "systemctl top"
2317 - "systemctl mask" should find all names by which a unit is accessible
2318 (i.e. by scanning for symlinks to it) and link them all to /dev/null
2321 - emulate /dev/kmsg using CUSE and turn off the syslog syscall
2322 with seccomp. That should provide us with a useful log buffer that
2323 systemd can log to during early boot, and disconnect container logs
2324 from the kernel's logs.
2325 - as soon as networkd has a bus interface, hook up --network-interface=,
2326 --network-bridge= with networkd, to trigger netdev creation should an
2327 interface be missing
2328 - a nice way to boot up without machine id set, so that it is set at boot
2329 automatically for supporting --ephemeral. Maybe hash the host machine id
2330 together with the machine name to generate the machine id for the container
2331 - fix logic always print a final newline on output.
2332 https://github.com/systemd/systemd/pull/272#issuecomment-113153176
2333 - should optionally support receiving WATCHDOG=1 messages from its payload
2335 - optionally automatically add FORWARD rules to iptables whenever nspawn is
2336 running, remove them when shut down.
2337 - add support for sysext extensions, too. i.e. a new --extension= switch that
2338 takes one or more arguments, and applies the extensions already during
2340 - when main nspawn supervisor process gets suspended due to SIGSTOP/SIGTTOU
2341 or so, freeze the payload too.
2342 - support time namespaces
2343 - on cgroupsv1 issue cgroup empty handler process based on host events, so
2344 that we make cgroup agent logic safe
2345 - add API to invoke binary in container, then use that as fallback in
2347 - make nspawn suitable for shell pipelines: instead of triggering a hangup
2348 when input is finished, send ^D, which synthesizes an EOF. Then wait for
2349 hangup or ^D before passing on the EOF.
2350 - greater control over selinux label?
2351 - support that /proc, /sys/, /dev are pre-mounted
2352 - maybe allow TPM passthrough, backed by swtpm, and measure --image= hash
2353 into its PCR 11, so that nspawn instances can be TPM enabled, and partake
2354 in measurements/remote attestation and such. swtpm would run outside of
2355 control of container, and ideally would itself bind its encryption keys to
2357 - make boot assessment do something sensible in a container. i.e send an
2358 sd_notify() from payload to container manager once boot-up is completed
2359 successfully, and use that in nspawn for dealing with boot counting,
2360 implemented in the partition table labels and directory names.
2361 - optionally set up nftables/iptables routes that forward UDP/TCP traffic on
2362 port 53 to resolved stub 127.0.0.54
2363 - maybe optionally insert .nspawn file as GPT partition into images, so that
2364 such container images are entirely stand-alone and can be updated as one.
2365 - The subreaper logic we currently have seems overly complex. We should
2366 investigate whether creating the inner child with CLONE_PARENT isn't better.
2367 - Reduce the number of sockets that are currently in use and just rely on one
2369 - Support running nspawn as an unprivileged user.
2371 * machined: add API to acquire UID range. add API to mount/dissect loopback
2372 file. Both protected by PK. Then make nspawn use these APIs to run
2373 unprivileged containers. i.e. push the truly privileged bits into machined,
2374 so that the client side can remain entirely unprivileged, with SUID or
2378 - add an API so that libvirt-lxc can inform us about network interfaces being
2379 removed or added to an existing machine
2380 - "machinectl migrate" or similar to copy a container from or to a
2381 difference host, via ssh
2382 - introduce systemd-nspawn-ephemeral@.service, and hook it into
2383 "machinectl start" with a new --ephemeral switch
2384 - "machinectl status" should also show internal logs of the container in
2386 - "machinectl history"
2388 - "machinectl commit" that takes a writable snapshot of a tree, invokes a
2389 shell in it, and marks it read-only after use
2394 - add trigger --subsystem-match=usb/usb_device device
2395 - reimport udev db after MOVE events for devices without dev_t
2396 - re-enable ProtectClock= once only cgroupsv2 is supported.
2397 See f562abe2963bad241d34e0b308e48cf114672c84.
2400 - save coredump in Windows/Mozilla minidump format
2401 - when truncating coredumps, also log the full size that the process had, and make a metadata field so we can report truncated coredumps
2402 - add examples for other distros in ELF_PACKAGE_METADATA
2404 * support crash reporting operation modes (https://live.gnome.org/GnomeOS/Design/Whiteboards/ProblemReporting)
2407 - apply "x" on "D" too (see patch from William Douglas)
2408 - allow time-based cleanup in r and R too
2409 - instead of ignoring unknown fields, reject them.
2410 - creating new directories/subvolumes/fifos/device nodes
2411 should not follow symlinks. None of the other adjustment or creation
2412 calls follow symlinks.
2414 - teach tmpfiles.d q/Q logic something sensible in the context of XFS/ext4
2416 - teach tmpfiles.d m/M to move / atomic move + symlink old -> new
2417 - add new line type for setting btrfs subvolume attributes (i.e. rw/ro)
2418 - tmpfiles: add new line type for setting fcaps
2421 - Make sure ID_PATH is always exported and complete for
2422 network devices where possible, so we can safely rely
2426 - add support for more attribute types
2427 - inbuilt piping support (essentially degenerate async)? see loopback-setup.c and other places
2430 - add more keys to [Route] and [Address] sections
2431 - add support for more DHCPv4 options (and, longer term, other kinds of dynamic config)
2432 - add reduced [Link] support to .network files
2433 - properly handle routerless dhcp leases
2434 - work with non-Ethernet devices
2435 - dhcp: do we allow configuring dhcp routes on interfaces that are not the one we got the dhcp info from?
2436 - the DHCP lease data (such as NTP/DNS) is still made available when
2437 a carrier is lost on a link. It should be removed instantly.
2438 - expose in the API the following bits:
2439 - option 15, domain name
2440 - option 12, hostname and/or option 81, fqdn
2441 - option 123, 144, geolocation
2442 - option 252, configure http proxy (PAC/wpad)
2443 - provide a way to define a per-network interface default metric value
2444 for all routes to it. possibly a second default for DHCP routes.
2445 - allow Name= to be specified repeatedly in the [Match] section. Maybe also
2446 support Name=foo*|bar*|baz ?
2447 - whenever uplink info changes, make DHCP server send out FORCERENEW
2449 * in networkd, when matching device types, fix up DEVTYPE rubbish the kernel passes to us
2451 * Figure out how to do unittests of networkd's state serialization
2454 - figure out how much we can increase Maximum Message Size
2457 - add functions to set previously stored IPv6 addresses on startup and get
2458 them at shutdown; store them in client->ia_na
2459 - write more test cases
2460 - implement reconfigure support, see 5.3., 15.11. and 22.20.
2461 - implement support for temporary adressess (IA_TA)
2462 - implement dhcpv6 authentication
2463 - investigate the usefulness of Confirm messages; i.e. are there any
2464 situations where the link changes without any loss in carrier detection
2466 - some servers don't do rapid commit without a filled in IA_NA, verify