]>
Commit | Line | Data |
---|---|---|
afe3ab58 | 1 | Bugfixes: |
c343be28 | 2 | |
54fcb619 ZJS |
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. | |
6 | ||
5b326dee ZJS |
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. | |
10 | ||
f38afcd0 | 11 | External: |
f85857df | 12 | |
f38afcd0 | 13 | * Fedora: add an rpmlint check that verifies that all unit files in the RPM are listed in %systemd_post macros. |
bafb15ba | 14 | |
07eabc2b LB |
15 | * dbus: |
16 | - natively watch for dbus-*.service symlinks (PENDING) | |
17 | - teach dbus to activate all services it finds in /etc/systemd/services/org-*.service | |
18 | ||
19 | * kernel: add device_type = "fb", "fbcon" to class "graphics" | |
20 | ||
21 | * /usr/bin/service should actually show the new command line | |
22 | ||
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= | |
24 | ||
25 | * neither pkexec nor sudo initialize environ[] from the PAM environment? | |
26 | ||
27 | * fedora: update policy to declare access mode and ownership of unit files to root:root 0644, and add an rpmlint check for it | |
28 | ||
29 | * register catalog database signature as file magic | |
30 | ||
31 | * zsh shell completion: | |
32 | - <command> <verb> -<TAB> should complete options, but currently does not | |
33 | - systemctl add-wants,add-requires | |
76c068b7 | 34 | - systemctl reboot --boot-loader-entry= |
07eabc2b LB |
35 | |
36 | * systemctl status should know about 'systemd-analyze calendar ... --iterations=' | |
37 | * If timer has just OnInactiveSec=..., it should fire after a specified time | |
38 | after being started. | |
39 | ||
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 | |
44 | - remote access | |
45 | - how to pass throw-away units to systemd, or dynamically change properties of existing units | |
46 | - testing with Harald's awesome test kit | |
47 | - auto-restart | |
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 | |
58 | ||
59 | Regularly: | |
60 | ||
61 | * look for close() vs. close_nointr() vs. close_nointr_nofail() | |
62 | ||
63 | * check for strerror(r) instead of strerror(-r) | |
64 | ||
65 | * pahole | |
66 | ||
67 | * set_put(), hashmap_put() return values check. i.e. == 0 does not free()! | |
68 | ||
69 | * use secure_getenv() instead of getenv() where appropriate | |
70 | ||
71 | * link up selected blog stories from man pages and unit files Documentation= fields | |
72 | ||
5e524b40 LP |
73 | Janitorial Clean-ups: |
74 | ||
bb527e11 LP |
75 | * rework mount.c and swap.c to follow proper state enumeration/deserialization |
76 | semantics, like we do for device.c now | |
77 | ||
f461a28d | 78 | * get rid of prefix_roota() and similar, only use chase() and related |
180efdb7 LP |
79 | calls instead. |
80 | ||
81 | * get rid of basename() and replace by path_extract_filename() | |
82 | ||
ca424984 ZJS |
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. | |
86 | ||
f24e1b54 LP |
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. | |
90 | ||
33bc21c3 LP |
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. | |
95 | ||
3345802c LP |
96 | Deprecations and removals: |
97 | ||
98 | * Remove any support for booting without /usr pre-mounted in the initrd entirely. | |
99 | Update INITRD_INTERFACE.md accordingly. | |
100 | ||
b98445cd LP |
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. | |
105 | ||
3345802c LP |
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 | |
9eb41aab | 108 | is only supported for such old kernels. |
3345802c | 109 | |
f18b0a76 LP |
110 | * drop support for kernels lacking memfd_create() (i.e. make 3.17 new |
111 | baseline), then drop all pipe() based fallbacks. | |
112 | ||
9eb41aab | 113 | * drop support for getrandom()-less kernels. (GRND_INSECURE means once kernel |
3345802c | 114 | 5.6 becomes our baseline). See |
9eb41aab LP |
115 | https://github.com/systemd/systemd/pull/24101#issuecomment-1193966468 for |
116 | details. Maybe before that: at taint-flags/warn about kernels that lack | |
3345802c LP |
117 | getrandom()/environments where it is blocked. |
118 | ||
9eb41aab LP |
119 | * drop support for LOOP_CONFIGURE-less loopback block devices, once kernel |
120 | baseline is 5.8. | |
3345802c | 121 | |
dec6f619 LP |
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 | |
124 | ||
bc09c2d8 JL |
125 | * Remove /dev/mem ACPI FPDT parsing when /sys/firmware/acpi/fpdt is ubiquitous. |
126 | That requires distros to enable CONFIG_ACPI_FPDT, and have kernels v5.12 for | |
127 | x86 and v6.2 for arm. | |
128 | ||
43057bf6 LP |
129 | * Once baseline is 4.13, remove support for INTERFACE_OLD= checks in "udevadm |
130 | trigger"'s waiting logic, since we can then rely on uuid-tagged uevents | |
131 | ||
459d8407 LP |
132 | * remove remaining tpm1.2 support from sd-stub |
133 | ||
3345802c LP |
134 | Features: |
135 | ||
9f3ed6fd LP |
136 | * redefine /var/lib/extensions/ as the dir one can place all three of sysext, |
137 | confext as well is multi-modal DDIs that qualify as both. Then introduce | |
138 | /var/lib/sysexts/ which can be used to place only DDIs that shall be used as | |
139 | sysext | |
140 | ||
69feab97 LP |
141 | * in pid1: move out all cgroup state settings from Unit into a new object |
142 | CGroupState or so which is allocated when we realize the unit into a cgroup, | |
143 | and then remains referenced by it. The new object should also carry an fd to | |
144 | the realized cgroup, to pin it (and later execute all cgroup operations over, | |
145 | once we drop cgroupv1 compat). | |
146 | ||
b28bd482 LP |
147 | * add new "systemd-ssh-generator", which allows basic ssh config via |
148 | credentials (host key). It generates sshd.socket for IP, but also | |
149 | sshd-vsock.socket for listening on AF_VSOCK when running in a VM, and | |
150 | sshd-unix.socket on AF_UNIX when running in a container. It also generates a | |
151 | matching sshd.service file with a host key passed in on the cmdline via | |
152 | credentials. Then, add a ssh_config drop-in that matches some suitable | |
153 | hostname pattern and has a ProxyCommand set that allows connecting to any | |
154 | local VM/container that way without any networking configured. | |
155 | ||
3d1831b9 LP |
156 | * Varlinkification of the following command line tools, to open them up to |
157 | other programs via IPC: | |
158 | - bootctl | |
159 | - journalctl (allowing journal read access via IPC) | |
160 | - coredumpcl | |
161 | - systemd-bless-boot | |
162 | - systemd-measure | |
163 | - systemd-creds (allowing clients to encrypt credentials locally) | |
164 | - systemd-cryptenroll (to allow UIs to enroll FIDO2 keys and such) | |
165 | - systemd-dissect | |
166 | - systemd-sysext | |
167 | - systemd-sysupdate | |
168 | - systemd-analyze | |
169 | - kernel-install | |
170 | ||
171 | * Varlink: add glue code to allow varlink clients to be authenticated via | |
172 | Polkit by passing client pidfd over. | |
173 | ||
02c9d62b LP |
174 | * in the service manager, pick up ERRNO= + BUSERROR= + VARLINKERROR= error |
175 | identifiers, and store them along with the exit status of a server and report | |
176 | via "systemctl status". | |
177 | ||
90eabfe6 LP |
178 | * enumerate virtiofs devices during boot-up in a generator, and synthesize |
179 | mounts for rootfs, /usr/, /home/, /srv/ and some others from it, depending on | |
02c9d62b | 180 | the "tag". (waits for: https://gitlab.com/virtio-fs/virtiofsd/-/issues/128) |
90eabfe6 LP |
181 | |
182 | * automatically mount one virtiofs during early boot phase to /run/host/, | |
183 | similar to how we do that for nspawn, based on some clear tag. | |
184 | ||
8b9c3af0 LP |
185 | * add some service that makes an atomic snapshot of PCR state and event log up |
186 | to that point available, possibly even with quote by the TPM. | |
187 | ||
8f533730 LP |
188 | * encode type1 entries in some UKI section to add additional entries to the |
189 | menu. | |
190 | ||
3d1831b9 LP |
191 | * Add ACL-based access management to .socket units. i.e. add AllowPeerUser= + |
192 | AllowPeerGroup= that installs additional user/group ACL entries on AF_UNIX | |
193 | sockets. | |
8f533730 | 194 | |
1a9454a9 LP |
195 | * systemd-tpm2-setup should probably have a factory reset logic, i.e. when some |
196 | kernel command line option is set we reset the TPM (equivalent of tpm2_clear | |
197 | -c owner?). | |
198 | ||
199 | * systemd-tpm2-setup should support a mode where we refuse booting if the SRK | |
200 | changed. (Must be opt-in, to not break systems which are supposed to be | |
201 | migratable between PCs) | |
202 | ||
8f533730 LP |
203 | * when systemd-sysext learns mutable /usr/ (and systemd-confext mutable /etc/) |
204 | then allow them to store the result in a .v/ versioned subdir, for some basic | |
205 | snapshot logic | |
206 | ||
207 | * add a new PE binary section ".mokkeys" or so which sd-stub will insert into | |
208 | Mok keyring, by overriding/extending whatever shim sets in the EFI | |
209 | var. Benefit: we can extend the kernel module keyring at ukify time, | |
fcdd21ec | 210 | i.e. without recompiling the kernel, taking an upstream OS' kernel and adding |
8f533730 LP |
211 | a local key to it. |
212 | ||
a1f7cdc6 LP |
213 | * PidRef conversion work: |
214 | - pid_is_unwaited() → pidref_is_unwaited() | |
215 | - pid_is_alive() → pidref_is_alive() | |
d94dc50c LP |
216 | - is_kernel_thread() → pidref_is_kernel_thread() |
217 | - pid_is_my_child() → pidref_is_my_child() | |
d94dc50c | 218 | - get_process_uid() → pidref_get_uid() |
4775b55d | 219 | - actually wait for POLLIN on pidref's pidfd in service logic |
a1f7cdc6 | 220 | - exec_spawn() |
a1f7cdc6 LP |
221 | - cg_pid_get_unit() |
222 | - openpt_allocate_in_namespace() | |
d94dc50c LP |
223 | - sd_bus_creds |
224 | - unit_attach_pid_to_cgroup_via_bus() | |
225 | - cg_attach() – requires new kernel feature | |
226 | - varlink_get_peer_pid() | |
a1f7cdc6 | 227 | |
354e5b88 LP |
228 | * ddi must be listed as block device fstype |
229 | ||
230 | * measure some string via pcrphase whenever we end up booting into emergency | |
231 | mode. | |
232 | ||
233 | * homed: add a basic form of of secrets management to homed, that stores | |
234 | secrets in $HOME somewhere, is protected by the accounts own authentication | |
235 | mechanisms. Should implement something PKCS#11-like that can be used to | |
236 | implement emulated FIDO2 in unpriv userspace on top (which should happen | |
237 | outside of homed), emulated PKCS11, and libsecrets support. Operate with a | |
238 | 2nd key derived from volume key of the user, with which to wrap all | |
239 | keys. maintain keys in kernel keyring if possible. | |
240 | ||
97046f2f LP |
241 | * use sd-event ratelimit feature optionally for journal stream clients that log |
242 | too much | |
c5783453 | 243 | |
fcdd21ec | 244 | * systemd-mount should only consider modern file systems when mounting, similar |
c5783453 LP |
245 | to systemd-dissect |
246 | ||
54b8a816 LP |
247 | * new "systemd-pcrlock" component for dealing with PCR4. Design idea: |
248 | 1. define /{etc,usr,var/lib}/pcrlock.d/<component>/<version>.pcrlock | |
249 | 2. these files contain list of hashes that will be measured when component is | |
250 | run, per PCR | |
251 | 3. each component involved in the boot that is deterministically measured can | |
252 | place one or more of these files in those dirs (shim, sd-boot, | |
253 | sd-stub/UKI, cryptsetup, pcrphase, pcrfs, …) | |
254 | 4. since each component has its own dir, with multiple files in them, package | |
255 | such as kernels (of which there can be multiple installed at the same | |
256 | time) can be grouped together: only one of them is measured at a time. | |
257 | 5. whenever a new component is added or an old one removed, or the PCR lock | |
258 | shall be relaxed or tightened the systemd-pcrlock tool is invoked. | |
259 | 6. tool iterates through all these files, orders them alphabetically by | |
260 | component, then matches them up with current measurements (as per uefi | |
261 | event log), identifying by hash, accepting that the "beginning" of the | |
262 | measurements might not be recognizable. | |
263 | 7. Then calculates expected PCR values starting with the "unrecognized | |
264 | head" from the event log, then continuing with all of components | |
265 | defined via the .pcrlock files (but dropping out the "recognized tail" | |
266 | from the uefi event log). (This might mean combinatorial explosion, if | |
267 | there are multiple shims, multiple sd-boot, and so on.) | |
268 | 8. Generates a public/private key pair on the TPM | |
269 | 9. Generates a counter object in the TPM, with a policy that allows only | |
270 | one-by-one increase with signature policy by the public/private key pair. | |
271 | 10. now signs policies of all expected PCR values with the generated keypair, | |
272 | using all combinations of components defined in the .pcrlock files | |
273 | restricting it to the counter + 1. | |
274 | 11. locks down the keypair with a signed policy with its own public key | |
275 | 12. generates JSON file of all these policies with their signatures, drops | |
276 | them as singleton in ESP | |
277 | 13. increases the counter by one. | |
278 | 14. after boot sd-stub picks JSON up from ESP, passes it to userspace via | |
279 | .extra | |
280 | 15. JSON contained policies can now be used to unlock disk as well as the | |
281 | public/key itself for signing further policies, as well as increment for | |
282 | the counter | |
283 | 16. whenever any of the components above is added/removed new JSON file with | |
284 | signatures for counter + 1 is generated, dropped in ESP, then counter | |
285 | increased. (i.e. this means the "recognized tail" of the event log is | |
286 | deterministically swapped out) | |
287 | 17. when firmware update is expected, relaxed signed policy is generated for | |
288 | next boot only valid if counter is increased (this means the | |
289 | "unrecognized head" for the event log can change without losing access) | |
290 | 18. on every boot checks if releaxed policy is in effect, if so, new strict | |
291 | policy is generated and counter increased. | |
292 | Net result: Removes downgrade attack surface + Locks OS to firmware + Allows | |
293 | downgrades within bounds | |
294 | ||
be76bf62 LP |
295 | * add another PE section ".fname" or so that encodes the intended filename for |
296 | PE file, and validate that when loading add-ons and similar before using | |
297 | it. This is particularly relevant when we load multiple add-ons and want to | |
298 | sort them to apply them in a define order. The order should not be under | |
299 | control of the attacker. | |
300 | ||
c4ffb530 LP |
301 | * also include packaging metadata (á la |
302 | https://systemd.io/ELF_PACKAGE_METADATA/) in our UEFI PE binaries, using the | |
303 | same JSON format. | |
304 | ||
305 | * make "bootctl install" + "bootctl update" useful for installing shim too. For | |
306 | that introduce new dir /usr/lib/systemd/efi/extra/ which we copy mostly 1:1 | |
307 | into the ESP at install time. Then make the logic smart enough so that we | |
308 | don't overwrite bootx64.efi with our own if the extra tree already contains | |
309 | one. Also, follow symlinks when copying, so that shim rpm can symlink their | |
310 | stuff into our dir (which is safe since the target ESP is generally VFAT and | |
311 | thus does not have symlinks anyway). Later, teach the update logic to look at | |
312 | the ELF package metadata (which we also should include in all PE files, see | |
313 | above) for version info in all *.EFI files, and use it to only update if | |
314 | newer. | |
315 | ||
0f85a0d3 LP |
316 | * in sd-stub: optionally add support for a new PE section .keyring or so that |
317 | contains additional certificates to include in the Mok keyring, extending | |
318 | what shim might have placed there. why? let's say I use "ukify" to build + | |
319 | sign my own fedora-based UKIs, and only enroll my personal lennart key via | |
320 | shim. Then, I want to include the fedora keyring in it, so that kmods work. | |
321 | But I might not want to enroll the fedora key in shim, because this would | |
322 | also mean that the key would be in effect whenever I boot an archlinux UKI | |
323 | built the same way, signed with the same lennart key. | |
324 | ||
29de4eba LP |
325 | * resolved: take possession of some IPv6 ULA address (let's say |
326 | fd00:5353:5353:5353:5353:5353:5353:5353), and listen on port 53 on it for the | |
327 | local stubs, so that we can make the stub available via ipv6 too. | |
328 | ||
9c9b9b89 LP |
329 | * introduce a .microcode PE section for sd-stub which we'll pass as first initrd |
330 | to the kernel which will then upload it to the CPU. This should be distinct | |
331 | from .initrd to guarantee right ordering. also, and maybe more importantly | |
332 | support .microcode in PE add-ons, so that a microcode update can be shipped | |
9a27ef09 | 333 | independently of any kernel. |
9c9b9b89 | 334 | |
aafd429c LP |
335 | * Maybe add SwitchRootEx() as new bus call that takes env vars to set for new |
336 | PID 1 as argument. When adding SwitchRootEx() we should maybe also add a | |
337 | flags param that allows disabling and enabling whether serialization is | |
338 | requested during switch root. | |
9c9b9b89 | 339 | |
89ee9e59 LP |
340 | * introduce a .acpitable section for early ACPI table override |
341 | ||
9c9b9b89 LP |
342 | * add proper .osrel matching for PE addons. i.e. refuse applying an addon |
343 | intended for a different OS. Take inspiration from how confext/sysext are | |
344 | matched against OS. | |
345 | ||
5596fd8f LP |
346 | * figure out what to do about credentials sealed to PCRs in kexec + soft-reboot |
347 | scenarios. Maybe insist sealing is done additionally against some keypair in | |
348 | the TPM to which access is updated on each boot, for the next, or so? | |
349 | ||
9287925a LP |
350 | * logind: when logging in, always take an fd to the home dir, to keep the dir |
351 | busy, so that autofs release can never happen. (this is generally a good | |
352 | idea, and specifically works around the fact the autofs ignores busy by mount | |
353 | namespaces) | |
354 | ||
f351e951 LP |
355 | * mount most file systems with a restrictive uidmap. e.g. mount /usr/ with a |
356 | uidmap that blocks out anything outside 0…1000 (i.e. system users) and similar. | |
357 | ||
358 | * mount the root fs with MS_NOSUID by default, and then mount /usr/ without | |
359 | both so that suid executables can only be placed there. Do this already in | |
360 | the initrd. If /usr/ is not split out create a bind mount automatically. | |
361 | ||
f447b741 | 362 | * fix our various hwdb lookup keys to end with ":" again. The original idea was |
94d82b59 | 363 | that hwdb patterns can match arbitrary fields with expressions like |
f447b741 LP |
364 | "*:foobar:*", to wildcard match both the start and the end of the string. |
365 | This only works safely for later extensions of the string if the strings | |
366 | always end in a colon. This requires updating our udev rules, as well as | |
367 | checking if the various hwdb files are fine with that. | |
368 | ||
c1c4ecd3 LP |
369 | * mount /tmp/ and /var/tmp with a uidmap applied that blocks out "nobody" user |
370 | among other things such as dynamic uid ranges for containers and so on. That | |
94d82b59 | 371 | way no one can create files there with these uids and we enforce they are only |
c1c4ecd3 LP |
372 | used transiently, never persistently. |
373 | ||
aa03f499 LP |
374 | * rework loopback support in fstab: when "loop" option is used, then |
375 | instantiate a new systemd-loop@.service for the source path, set the | |
376 | lo_file_name field for it to something recognizable derived from the fstab | |
377 | line, and then generate a mount unit for it using a udev generated symlink | |
378 | based on lo_file_name. | |
379 | ||
a69a520b LP |
380 | * remove tomoyo support, it's obsolete and unmaintained apparently |
381 | ||
8b098d94 LP |
382 | * In .socket units, add ConnectStream=, ConnectDatagram=, |
383 | ConnectSequentialPacket= that create a socket, and then *connect to* rather than | |
384 | listen on some socket. Then, add a new setting WriteData= that takes some | |
385 | base64 data that systemd will write into the socket early on. This can then | |
386 | be used to create connections to arbitrary services and issue requests into | |
387 | them, as long as the data is static. This can then be combined with the | |
388 | aforementioned journald subscription varlink service, to enable | |
389 | activation-by-message id and similar. | |
390 | ||
ed856ea1 ZJS |
391 | * .service with invalid Sockets= starts successfully. |
392 | ||
512f2da5 LP |
393 | * landlock: lock down RuntimeDirectory= via landlock, so that services lose |
394 | ability to write anywehere else below /run/. Similar for | |
395 | StateDirectory=. Benefit would be clear delegation via unit files: services | |
396 | get the directories they get, and nothing else even if they wanted to. | |
397 | ||
398 | * landlock: for unprivileged systemd (i.e. systemd --user), use landlock to | |
399 | implement ProtectSystem=, ProtectHome= and so on. Landlock does not require | |
400 | privs, and we can implement pretty similar behaviour. Also, maybe add a mode | |
401 | where ProtectSystem= combined with an explicit PrivateMounts=no could request | |
402 | similar behaviour for system services, too. | |
403 | ||
13340701 LP |
404 | * Add systemd-mount@.service which is instantiated for a block device and |
405 | invokes systemd-mount and exits. This is then useful to use in | |
406 | ENV{SYSTEMD_WANTS} in udev rules, and a bit prettier than using RUN+= | |
407 | ||
3d092a70 LP |
408 | * udevd: extend memory pressure logic: also kill any idle worker processes |
409 | ||
12c346d8 ZJS |
410 | * udevadm: to make symlink querying with udevadm nicer: |
411 | - do not enable the pager for queries like 'udevadm info -q -r symlink' | |
412 | - add mode with newlines instead of spaces (for grep)? | |
413 | ||
3d092a70 LP |
414 | * SIGRTMIN+18 and memory pressure handling should still be added to: hostnamed, |
415 | localed, oomd, timedated. | |
416 | ||
1c904337 LP |
417 | * in order to make binding to PCR 4 realistic: |
418 | - generate one keypair "U" and store it in a tpm2 nvindex. | |
419 | - Generate another keypair "P" and store it in a second tpm2 nvindex. | |
420 | - allocate a persistent counter object "C" in the tpm2 | |
421 | - Enroll all user objects (i.e. luks volumes, creds, …) to a tpm2 policy | |
422 | signed by U. | |
423 | - Lock both U and P down with a tpm2 policy signed by P (yes, P can only be | |
424 | used if a signature by P itself can be provided) | |
425 | - For regular reboots generate a signature for a restrictive PCR4 + counter C | |
426 | based policy with key P. Place signature in EFI var, so it can be found on | |
427 | next boot | |
428 | - For reboots where a firmware update is expected generate a signature with a | |
429 | more open policy against just counter C. Place signature in same EFI var. | |
430 | - Increase C whenever switching between these two signature types. | |
431 | - During early boot, use the signature from the EFI var to unlock U and P. | |
432 | Use it to generate a signature for unlocking user objects given the current | |
433 | PCR 4 value, store that away into /run somewhere, for user during the whole | |
434 | later boot. | |
435 | - When booting up automatically update the mentioned efi var so that it | |
436 | contains the restrictive signature. But also generate a signature ahead of | |
437 | time that could be used in case during the current boot we later detect we might | |
438 | need to reboot for a firmware update. Store that in /run somewhere, so that | |
439 | it can be placed in the EFI var, if needed. | |
440 | ||
441 | * repart/gpt-auto/DDIs: maybe introduce a concept of "extension" partitions, | |
442 | that have a new type uuid and can "extend" earlier partitions, to work around | |
443 | the fact that systemd-repart can only grow the last partition defined. During | |
444 | activation we'd simply set up a dm-linear mapping to merge them again. A | |
445 | partition that is to be extended would just set a bit in the partition flags | |
446 | field to indicate that there's another extension partition to look for. The | |
627cdcc7 | 447 | identifying UUID of the extension partition would be hashed in counter mode |
1c904337 LP |
448 | from the uuid of the original partition it extends. Inspiration for this is |
449 | the "dynamic partitions" concept of new Android. This would be a minimalistic | |
450 | concept of a volume manager, with the extents it manages being exposes as GPT | |
451 | partitions. I a partition is extended multiple times they should probably | |
452 | grow exponentially in size to ensure O(log(n)) time for finding them on | |
453 | access. | |
8f8d7dff | 454 | |
32284ffc LP |
455 | * split out execute.c into new "systemd-executor" binary. Then make PID 1 fork |
456 | that off via vfork(), and then let that executor do the hard work. Ultimately | |
457 | the executor then gets replaced by the real binary sooner or later. Reason: | |
458 | currently the intermediary "stub" process is a CoW trap that doubles memory | |
459 | usage of PID 1 on each service start. Also, strictly speaking we are not | |
d09df6b9 | 460 | allowed to do NSS from the stub process yet we do anyway. Next steps would |
32284ffc LP |
461 | then be maybe use CLONE_INTO_CGROUP for the executor, given that we don't |
462 | need glibc anymore in the stub process then. Then, switch nspawn to just be a | |
463 | frontend for this too, so that we have to ways into the executor: via unit | |
464 | files/dbus/varlin through PID1 and via cmdline/OCI through nspawn. | |
465 | ||
46730cd6 LP |
466 | * sd-stub: detect if we are running with uefi console output on serial, and if so |
467 | automatically add console= to kernel cmdline matching the same port. | |
468 | ||
e793038a LP |
469 | * add a utility that can be used with the kernel's |
470 | CONFIG_STATIC_USERMODEHELPER_PATH and then handles them within pid1 so that | |
471 | security, resource management and cgroup settings can be enforced properly | |
472 | for all umh processes. | |
473 | ||
f912cdbc LP |
474 | * systemd-shutdown: keep sending sd_notify() status updates immediately before |
475 | going down, in particular include the "reboot param" string. | |
476 | ||
a1ad873d LP |
477 | * homed: when resizing an fs don't sync identity beforehand there might simply |
478 | not be enough disk space for that. try to be defensive and sync only after | |
479 | resize. | |
480 | ||
481 | * homed: if for some reason the partition ended up being much smaller than | |
482 | whole disk, recover from that, and grow it again. | |
483 | ||
a0bccdd3 LP |
484 | * timesyncd: when saving/restoring clock try to take boot time into account. |
485 | Specifically, along with the saved clock, store the current boot ID. When | |
486 | starting, check if the boot id matches. If so, don't do anything (we are on | |
487 | the same boot and clock just kept running anyway). If not, then read | |
488 | CLOCK_BOOTTIME (which started at boot), and add it to the saved clock | |
489 | timestamp, to compensate for the time we spent booting. If EFI timestamps are | |
490 | available, also include that in the calculation. With this we'll then only | |
491 | miss the time spent during shutdown after timesync stopped and before the | |
492 | system actually reset. | |
493 | ||
494 | * systemd-stub: maybe store a "boot counter" in the ESP, and pass it down to | |
e591cd5c LP |
495 | userspace to allow ordering boots (for example in journalctl). The counter |
496 | would be monotonically increased on every boot. | |
a0bccdd3 | 497 | |
87a7084b LP |
498 | * pam_systemd_home: add module parameter to control whether to only accept |
499 | only password or only pcks11/fido2 auth, and then use this to hook nicely | |
500 | into two of the three PAM stacks gdm provides. | |
501 | See discussion at https://github.com/authselect/authselect/pull/311 | |
502 | ||
4cb8a25b | 503 | * sd-boot: make boot loader spec type #1 accept http urls in "linux" |
1c904337 | 504 | lines. Then, do the uefi http dance to download kernels and boot them. This |
fcdd21ec | 505 | is then useful for network boot, by embedding a cpio with type #1 snippets |
4cb8a25b LP |
506 | in sd-boot, which reference remote kernels. |
507 | ||
d995ccd7 LP |
508 | * maybe prohibit setuid() to the nobody user, to lock things down, via seccomp. |
509 | the nobody is not a user any code should run under, ever, as that user would | |
510 | possibly get a lot of access to resources it really shouldn't be getting | |
511 | access to due to the userns + nfs semantics of the user. Alternatively: use | |
512 | the seccomp log action, and allow it. | |
513 | ||
0b556555 LP |
514 | * sd-boot: add a new PE section .bls or so that carries a cpio with additional |
515 | boot loader entries (both type1 and type2). Then when initializing, find this | |
516 | section, iterate through it and populate menu with it. cpio is simple enough | |
517 | to make a parser for this reasonably robust. use same path structures as in | |
518 | the ESP. Similar add one for signature key drop-ins. | |
519 | ||
4cb8a25b LP |
520 | * sd-boot: also allow passing in the cpio as in the previous item via SMBIOS |
521 | ||
0b556555 LP |
522 | * add a new EFI tool "sd-fetch" or so. It looks in a PE section ".url" for an |
523 | URL, then downloads the file from it using UEFI HTTP APIs, and executes it. | |
7227dd81 | 524 | Use case: provide a minimal ESP with sd-boot and a couple of these sd-fetch |
0b556555 LP |
525 | binaries in place of UKIs, and download them on-the-fly. |
526 | ||
084ecae5 | 527 | * maybe: systemd-loop-generator that sets up loopback devices if requested via kernel |
7227dd81 | 528 | cmdline. use case: include encrypted/verity root fs in UKI. |
084ecae5 LP |
529 | |
530 | * systemd-gpt-auto-generator: add kernel cmdline option to override block | |
531 | device to dissect. also support dissecting a regular file. useccase: include | |
532 | encrypted/verity root fs in UKI. | |
533 | ||
6d040d84 | 534 | * sd-stub: add ".bootcfg" section for kernel bootconfig data (as per |
4cb8a25b | 535 | https://docs.kernel.org/admin-guide/bootconfig.html) |
6d040d84 LP |
536 | |
537 | * tpm2: add (optional) support for generating a local signing key from PCR 15 | |
538 | state. use private key part to sign PCR 7+14 policies. stash signatures for | |
539 | expected PCR7+14 policies in EFI var. use public key part in disk encryption. | |
540 | generate new sigs whenever db/dbx/mok/mokx gets updated. that way we can | |
541 | securely bind against SecureBoot/shim state, without having to renroll | |
542 | everything on each update (but we still have to generate one sig on each | |
543 | update, but that should be robust/idempotent). needs rollback protection, as | |
544 | usual. | |
545 | ||
e8cb96ac LP |
546 | * Lennart: big blog story about DDIs |
547 | ||
548 | * Lennart: big blog story about building initrds | |
549 | ||
550 | * Lennart: big blog story about "why systemd-boot" | |
551 | ||
0cdb1a2f LP |
552 | * bpf: see if we can use BPF to solve the syslog message cgroup source problem: |
553 | one idea would be to patch source sockaddr of all AF_UNIX/SOCK_DGRAM to | |
554 | implicitly contain the source cgroup id. Another idea would be to patch | |
555 | sendto()/connect()/sendmsg() sockaddr on-the-fly to use a different target | |
556 | sockaddr. | |
557 | ||
558 | * bpf: see if we can address opportunistic inode sharing of immutable fs images | |
559 | with BPF. i.e. if bpf gives us power to hook into openat() and return a | |
560 | different inode than is requested for which we however it has same contents | |
561 | then we can use that to implement opportunistic inode sharing among DDIs: | |
562 | make all DDIs ship xattr on all reg files with a SHA256 hash. Then, also | |
563 | dictate that DDIs should come with a top-level subdir where all reg files are | |
564 | linked into by their SHA256 sum. Then, whenever an inode is opened with the | |
565 | xattr set, check bpf table to find dirs with hashes for other prior DDIs and | |
566 | try to use inode from there. | |
567 | ||
0cdb1a2f LP |
568 | * extend the verity signature partition to permit multiple signatures for the |
569 | same root hash, so that people can sign a single image with multiple keys. | |
570 | ||
efe7ab96 LP |
571 | * consider adding a new partition type, just for /opt/ for usage in system |
572 | extensions | |
573 | ||
574 | * gpt-auto-discovery: also use the pkcs7 signature stuff, and pass signature to | |
575 | kernel. So far we only did this for the various --image= switches, but not | |
576 | for the root fs or /usr/. | |
577 | ||
9ef6330e LP |
578 | * dissection policy should enforce that unlocking can only take place by |
579 | certain means, i.e. only via pw, only via tpm2, or only via fido, or a | |
580 | combination thereof. | |
581 | ||
582 | * make the systemd-repart "seed" value provisionable via credentials, so that | |
583 | confidential computing environments can set it and deterministically | |
584 | enforce the uuids for partitions created, so that they can calculate PCR 15 | |
585 | ahead of time. | |
586 | ||
587 | * systemd-repart: also derive the volume key from the seed value, for the | |
588 | aforementioned purpose. | |
589 | ||
590 | * in the initrd: derive the default machine ID to pass to the host PID 1 via | |
591 | $machine_id from the same seed credential. | |
592 | ||
f87338fa DDM |
593 | * Add systemd-sysupdate-initrd.service or so that runs systemd-sysupdate in the |
594 | initrd to bootstrap the initrd to populate the initial partitions. Some things | |
595 | to figure out: | |
596 | - Should it run on firstboot or on every boot? | |
597 | - If run on every boot, should it use the sysupdate config from the host on | |
598 | subsequent boots? | |
599 | ||
dbc62e59 | 600 | * provide an API (probably IPC) to apps to encrypt/decrypt |
7227dd81 | 601 | credentials. use case: allow bluez bluetooth daemon to pass pairings to initrd |
dbc62e59 | 602 | that way, without shelling out to our tools. |
c868e95e | 603 | |
924a329a LP |
604 | * revisit default PCR bindings in cryptenroll and systemd-creds. Currently they |
605 | use PCR 7 which should contain secureboot state db/dbx. Which sounded like a | |
606 | safe bet, given that it should change only on policy changes, and not | |
607 | software updates. But that's wrong. Recent fwupd (rightfully) contains code | |
608 | for updating the dbx denylist. This means even without any active policy | |
609 | change PCR 7 might change. Hence, better idea might be in systemd-creds to | |
6d040d84 LP |
610 | default to PCR 15 at least if sd-stub is used (i.e. bind to system identity), |
611 | and in cryptsetup simply the empty list? Also, PCR 14 almost certainly should | |
612 | be included as much as PCR 7 (as it contains shim's policy, which is | |
613 | certainly as relevant as PCR 7 on many systems) | |
924a329a | 614 | |
924a329a LP |
615 | * To mimic the new tpm2-measure-pcr= crypttab option add the same to veritytab |
616 | (measuring the root hash) and integritytab (measuring the HMAC key if one is | |
617 | used) | |
618 | ||
4554c178 LP |
619 | * We should start measuring all services, containers, and system extensions we |
620 | activate. probably into PCR 13. i.e. add --tpm2-measure-pcr= or so to | |
621 | systemd-nspawn, and MeasurePCR= to unit files. Should contain a measurement | |
622 | of the activated configuration and the image that is being activated (in case | |
623 | verity is used, hash of the root hash). | |
624 | ||
4d727f86 LP |
625 | * bootspec: permit graceful "update" from type #2 to type #1. If both a type #1 |
626 | and a type #2 entry exist under otherwise the exact same name, then use the | |
627 | type #1 entry, and ignore the type #2 entry. This way, people can "upgrade" | |
628 | from the UKI with all parameters baked in to a Type #1 .conf file with manual | |
7ff7eadf | 629 | parametrization, if needed. This matches our usual rule that admin config |
4d727f86 LP |
630 | should win over vendor defaults. |
631 | ||
4d727f86 LP |
632 | * write a "search path" spec, that documents the prefixes to search in |
633 | (i.e. the usual /etc/, /run/, /usr/lib/ dance, potentially /usr/etc/), how to | |
634 | sort found entries, how masking works and overriding. | |
635 | ||
636 | * automatic boot assessment: add one more default success check that just waits | |
637 | for a bit after boot, and blesses the boot if the system stayed up that long. | |
638 | ||
639 | * implement concept of "versioned" resources inside a dir, and write a spec for | |
640 | it. Make all tools in systemd, in particular | |
641 | RootImage=/RootDirectory=/--image=/--directory= implement this. Idea: | |
642 | directories ending in ".v/" indicate a directory with versioned resources in | |
643 | them. Versioned resources inside a .v dir are always named in the pattern | |
644 | <prefix>_<version>[+<tries-left>[-<tries-done>]].<suffix> | |
645 | ||
646 | * add support for using this .v/ logic on the root fs itself: in the initrd, | |
647 | after mounting the rootfs, look for root-<arch>.v/ in the root fs, and then | |
648 | apply the logic, moving the switch root logic there. | |
649 | ||
650 | * systemd-repart: add support for generating ISO9660 images | |
651 | ||
652 | * systemd-repart: in addition to the existing "factory reset" mode (which | |
653 | simply empties existing partitions marked for that). add a mode where | |
7227dd81 | 654 | partitions marked for it are entirely removed. Use case: remove secondary OS |
4d727f86 LP |
655 | copy, and redundant partitions entirely, and recreate them anew. |
656 | ||
657 | * systemd-boot: maybe add support for collapsing menu entries of the same OS | |
658 | into one item that can be opened (like in a "tree view" UI element) or | |
659 | collapsed. If only a single OS is installed, disable this mode, but if | |
660 | multiple OSes are installed might make sense to default to it, so that user | |
661 | is not immediately bombarded with a multitude of Linux kernel versions but | |
662 | only one for each OS. | |
663 | ||
664 | * systemd-repart: if the GPT *disk* UUID (i.e. the one global for the entire | |
665 | disk) is set to all FFFFF then use this as trigger for factory reset, in | |
7ff7eadf | 666 | addition to the existing mechanisms via EFI variables and kernel command |
4d727f86 LP |
667 | line. Benefit: works also on non-EFI systems, and can be requested on one |
668 | boot, for the next. | |
669 | ||
4d727f86 LP |
670 | * systemd-sysupdate: make transport pluggable, so people can plug casync or |
671 | similar behind it, instead of http. | |
672 | ||
673 | * systemd-tmpfiles: add concept for conditionalizing lines on factory reset | |
674 | boot, or on first boot. | |
675 | ||
676 | * in UKIs: add way to define allowlist of additional words that can be added to | |
677 | the kernel cmdline even in SecureBoot mode | |
678 | ||
a5a0da08 LP |
679 | * we probably needs .pcrpkeyrd or so as additional PE section in UKIs, |
680 | which contains a separate public key for PCR values that only apply in the | |
681 | initrd, i.e. in the boot phase "enter-initrd". Then, consumers in userspace | |
682 | can easily bind resources to just the initrd. Similar, maybe one more for | |
683 | "enter-initrd:leave-initrd" for resources that shall be accessible only | |
684 | before unprivileged user code is allowed. (we only need this for .pcrpkey, | |
685 | not for .pcrsig, since the latter is a list of signatures anyway). With that, | |
686 | when you enroll a LUKS volume or similar, pick either the .pcrkey (for | |
687 | coverage through all phases of the boot, but excluding shutdown), the | |
688 | .pcrpkeyrd (for coverage in the initrd only) and .pcrpkeybt (for coverage | |
689 | until users are allowed to log in). | |
690 | ||
691 | * Once the root fs LUKS volume key is measured into PCR 15, default to binding | |
692 | credentials to PCR 15 in "systemd-creds" | |
693 | ||
feffee70 LP |
694 | * add support for asymmetric LUKS2 TPM based encryption. i.e. allow preparing |
695 | an encrypted image on some host given a public key belonging to a specific | |
696 | other host, so that only hosts possessing the private key in the TPM2 chip | |
7227dd81 | 697 | can decrypt the volume key and activate the volume. Use case: systemd-confext |
b60e0f57 | 698 | for a central orchestrator to generate confext images securely that can only |
feffee70 LP |
699 | be activated on one specific host (which can be used for installing a bunch |
700 | of creds in /etc/credstore/ for example). Extending on this: allow binding | |
701 | LUKS2 TPM based encryption also to the TPM2 internal clock. Net result: | |
b60e0f57 | 702 | prepare a confext image that can only be activated on a specific host that |
703 | runs a specific software in a specific time window. confext would be | |
feffee70 LP |
704 | automatically invalidated outside of it. |
705 | ||
706 | * maybe add a "systemd-report" tool, that generates a TPM2-backed "report" of | |
707 | current system state, i.e. a combination of PCR information, local system | |
708 | time and TPM clock, running services, recent high-priority log | |
709 | messages/coredumps, system load/PSI, signed by the local TPM chip, to form an | |
7227dd81 | 710 | enhanced remote attestation quote. Use case: a simple orchestrator could use |
ae24e4e8 | 711 | this: have the report tool upload these reports every 3min somewhere. Then |
feffee70 LP |
712 | have the orchestrator collect these reports centrally over a 3min time |
713 | window, and use them to determine what which node should now start/stop what, | |
b60e0f57 | 714 | and generate a small confext for each node, that uses Uphold= to pin services |
715 | on each node. The confext would be encrypted using the asymmetric encryption | |
feffee70 LP |
716 | proposed above, so that it can only be activated on the specific host, if the |
717 | software is in a good state, and within a specific time frame. Then run a | |
718 | loop on each node that sends report to orchestrator and then sysupdate to | |
b60e0f57 | 719 | update confext. Orchestrator would be stateless, i.e. operate on desired |
feffee70 LP |
720 | config and collected reports in the last 3min time window only, and thus can |
721 | be trivially scaled up since all instances of the orchestrator should come to | |
722 | the same conclusions given the same inputs of reports/desired workload info. | |
723 | Could also be used to deliver Wireguard secrets and thus to clients, thus | |
b60e0f57 | 724 | permitting zero-trust networking: secrets are rolled over via confext updates, |
feffee70 LP |
725 | and via the time window TPM logic invalidated if node doesn't keep itself |
726 | updated, or becomes corrupted in some way. | |
727 | ||
a67a50e8 LP |
728 | * in the initrd, once the rootfs encryption key has been measured to PCR 15, |
729 | derive default machine ID to use from it, and pass it to host PID 1. | |
feffee70 | 730 | |
de76643b LP |
731 | * tree-wide: convert as much as possible over to use sd_event_set_signal_exit(), instead |
732 | of manually hooking into SIGINT/SIGTERM | |
733 | ||
734 | * tree-wide: convert as much as possible over to SD_EVENT_SIGNAL_PROCMASK | |
735 | instead of manual blocking. | |
736 | ||
eb8817db LP |
737 | * sd-boot: for each installed OS, grey out older entries (i.e. all but the |
738 | newest), to indicate they are obsolete | |
739 | ||
d1666bde LP |
740 | * automatically propagate LUKS password credential into cryptsetup from host |
741 | (i.e. SMBIOS type #11, …), so that one can unlock LUKS via VM hypervisor | |
742 | supplied password. | |
7dad7811 | 743 | |
72a77377 LP |
744 | * add ability to path_is_valid() to classify paths that refer to a dir from |
745 | those which may refer to anything, and use that in various places to filter | |
746 | early. i.e. stuff ending in "/", "/." and "/.." definitely refers to a | |
747 | directory, and paths ending that way can be refused early in many contexts. | |
748 | ||
fdcc31b7 LP |
749 | * systemd-measure: allow operating with PEM certificates in addition to PEM |
750 | public keys when signing PCR values. SecureBoot and our Verity signatures | |
751 | operate with certificates already, hence I guess we should also just deal for | |
fcdd21ec | 752 | convenience with certificates for the PCR stuff too. |
fdcc31b7 LP |
753 | |
754 | * systemd-measure: add --pcrpkey-auto as an alternative to --pcrpkey=, where it | |
755 | would just use the same public key specified with --public-key= (or the one | |
756 | automatically derived from --private-key=). | |
757 | ||
72a77377 LP |
758 | * push people to use ".sysext.raw" as suffix for sysext DDIs (DDI = |
759 | discoverable disk images, i.e. the new name for gpt disk images following the | |
760 | discoverable disk spec). [Also: just ".sysext/" for directory-based sysext] | |
761 | ||
762 | * Add "purpose" flag to partition flags in discoverable partition spec that | |
763 | indicate if partition is intended for sysext, for portable service, for | |
764 | booting and so on. Then, when dissecting DDI allow specifying a purpose to | |
7227dd81 | 765 | use as additional search condition. Use case: images that combined a sysext |
72a77377 LP |
766 | partition with a portable service partition in one. |
767 | ||
768 | * On boot, auto-generate an asymmetric key pair from the TPM, | |
769 | and use it for validating DDIs and credentials. Maybe upload it to the kernel | |
770 | keyring, so that the kernel does this validation for us for verity and kernel | |
771 | modules | |
772 | ||
b60e0f57 | 773 | * for systemd-confext: add a tool that can generate suitable DDIs with verity + |
72a77377 LP |
774 | sig using squashfs-tools-ng's library. Maybe just systemd-repart called under |
775 | a new name with a built-in config? | |
776 | ||
4e0ceefe LP |
777 | * lock down acceptable encrypted credentials at boot, via simple allowlist, |
778 | maybe on kernel command line: | |
779 | systemd.import_encrypted_creds=foobar.waldo,tmpfiles.extra to protect locked | |
780 | down kernels from credentials generated on the host with a weak kernel | |
781 | ||
4e0ceefe LP |
782 | * Add support for extra verity configuration options to systemd-repart (FEC, |
783 | hash type, etc) | |
0fc40a0e | 784 | |
f461a28d | 785 | * chase(): take inspiration from path_extract_filename() and return |
be429c8f LP |
786 | O_DIRECTORY if input path contains trailing slash. |
787 | ||
f461a28d | 788 | * chase(): refuse resolution if trailing slash is specified on input, |
4e0ceefe | 789 | but final node is not a directory |
be429c8f | 790 | |
be429c8f LP |
791 | * document in boot loader spec that symlinks in XBOOTLDR/ESP are not OK even if |
792 | non-VFAT fs is used. | |
793 | ||
a9f1bf40 LP |
794 | * measure credentials picked up from SMBIOS to some suitable PCR |
795 | ||
b7b7441d LP |
796 | * measure GPT and LUKS headers somewhere when we use them (i.e. in |
797 | systemd-gpt-auto-generator/systemd-repart and in systemd-cryptsetup?) | |
798 | ||
a9f1bf40 LP |
799 | * pick up creds from EFI vars |
800 | ||
acbb504e WR |
801 | * Add and pickup tpm2 metadata for creds structure. |
802 | ||
c794e280 LP |
803 | * sd-boot: we probably should include all BootXY EFI variable defined boot |
804 | entries in our menu, and then suppress ourselves. Benefit: instant | |
805 | compatibility with all other OSes which register things there, in particular | |
806 | on other disks. Always boot into them via NextBoot EFI variable, to not | |
807 | affect PCR values. | |
808 | ||
c0a74f62 LP |
809 | * systemd-measure tool: |
810 | - pre-calculate PCR 12 (command line) + PCR 13 (sysext) the same way we can precalculate PCR 11 | |
c0a74f62 | 811 | |
5b6e5d57 LP |
812 | * in sd-boot: load EFI drivers from a new PE section. That way, one can have a |
813 | "supercharged" sd-boot binary, that could carry ext4 drivers built-in. | |
814 | ||
1160267a LP |
815 | * sd-bus: document that sd_bus_process() only returns messages that non of the |
816 | filters/handlers installed on the connection took possession of. | |
817 | ||
812a8731 LP |
818 | * sd-device: add an API for acquiring list of child devices, given a device |
819 | objects (i.e. all child dirents that dirs or symlinks to dirs) | |
820 | ||
821 | * sd-device: maybe pin the sysfs dir with an fd, during the entire runtime of | |
822 | an sd_device, then always work based on that. | |
823 | ||
9f3a3ac7 | 824 | * add small wrapper around qemu that implements sd_notify/AF_VSOCK + machined and |
1fa5f1c5 LP |
825 | maybe some other stuff and boots it. Should implement command line roughly |
826 | equivalent to nspawn's. Maybe be called "systemd-vmspawn". Should imply good | |
827 | settings, i.e. RNG + HyperV enlightenments. Should also result in swtpm | |
828 | instance, plus virtiofsd instances. Translate credentials into smbios type | |
829 | 11 strings. Correctly translate SIGTERM into ACPI shutdown events. | |
830 | Listen to logind suspend events and turn these into suspend key pressed + | |
831 | ACPI resume events. | |
9f3a3ac7 | 832 | |
e1b45a75 | 833 | * maybe add new flags to gpt partition tables for rootfs and usrfs indicating |
9f3a3ac7 LP |
834 | purpose, i.e. whether something is supposed to be bootable in a VM, on |
835 | baremetal, on an nspawn-style container, if it is a portable service image, | |
836 | or a sysext for initrd, for host os, or for portable container. Then hook | |
837 | portabled/… up to udev to watch block devices coming up with the flags set, and | |
838 | use it. | |
839 | ||
b89cfe8a LP |
840 | * sd-boot should look for information what to boot in SMBIOS, too, so that VM |
841 | managers can tell sd-boot what to boot into and suchlike | |
842 | ||
52cd58b8 LP |
843 | * add "systemd-sysext identify" verb, that you can point on any file in /usr/ |
844 | and that determines from which overlayfs layer it originates, which image, and with | |
845 | what it was signed. | |
846 | ||
0fde330d LP |
847 | * systemd-creds: extend encryption logic to support asymmetric |
848 | encryption/authentication. Idea: add new verb "systemd-creds public-key" | |
849 | which generates a priv/pub key pair on the TPM2 and stores the priv key | |
850 | locally in /var. It then outputs a certificate for the pub part to stdout. | |
851 | This can then be copied/taken elsewhere, and can be used for encrypting creds | |
852 | that only the host on its specific hw can decrypt. Then, support a drop-in | |
853 | dir with certificates that can be used to authenticate credentials. Flow of | |
854 | operations is then this: build image with owner certificate, then after | |
855 | boot up issue "systemd-creds public-key" to acquire pubkey of the machine. | |
856 | Then, when passing data to the machine, sign with privkey belonging to one of | |
857 | the dropped in certs and encrypted with machine pubkey, and pass to machine. | |
858 | Machine is then able to authenticate you, and confidentiality is guaranteed. | |
859 | ||
bbe29ca2 LP |
860 | * building on top of the above, the pub/priv key pair generated on the TPM2 |
861 | should probably also one you can use to get a remote attestation quote. | |
862 | ||
d1666bde LP |
863 | * Process credentials in: |
864 | • networkd/udevd: add a way to define additional .link, .network, .netdev files | |
865 | via the credentials logic. | |
aafd429c LP |
866 | • crypttab-generator: allow defining additional crypttab-like volumes via |
867 | credentials (similar: verity-generator, integrity-generator). Use | |
868 | fstab-generator logic as inspiration. | |
d1666bde LP |
869 | • run-generator: allow defining additional commands to run via a credential |
870 | • resolved: allow defining additional /etc/hosts entries via a credential (it | |
871 | might make sense to then synthesize a new combined /etc/hosts file in /run | |
872 | and bind mount it on /etc/hosts for other clients that want to read it. | |
d1666bde LP |
873 | • repart: allow defining additional partitions via credential |
874 | • timesyncd: pick NTP server info from credential | |
875 | • portabled: read a credential "portable.extra" or so, that takes a list of | |
876 | file system paths to enable on start. | |
877 | • make systemd-fstab-generator look for a system credential encoding root= or | |
878 | usr= | |
879 | • systemd-homed: when initializing, look for a credential | |
880 | systemd.homed.register or so with JSON user records to automatically | |
7227dd81 | 881 | register if not registered yet. Use case: deploy a system, and add an |
d1666bde | 882 | account one can directly log into. |
d1666bde LP |
883 | • in gpt-auto-generator: check partition uuids against such uuids supplied via |
884 | sd-stub credentials. That way, we can support parallel OS installations with | |
885 | pre-built kernels. | |
bbe29ca2 | 886 | |
f95db4d6 LP |
887 | * define a JSON format for units, separating out unit definitions from unit |
888 | runtime state. Then, expose it: | |
889 | ||
890 | 1. Add Describe() method to Unit D-Bus object that returns a JSON object | |
891 | about the unit. | |
892 | 2. Expose this natively via Varlink, in similar style | |
893 | 3. Use it when invoking binaries (i.e. make PID 1 fork off systemd-executor | |
894 | binary which reads the JSON definition and runs it), to address the cow | |
895 | trap issue and the fact that NSS is actually forbidden in | |
896 | forked-but-not-exec'ed children | |
897 | 4. Add varlink API to run transient units based on provided JSON definitions | |
898 | ||
81a96518 LP |
899 | * Add SUPPORT_END_URL= field to os-release with more *actionable* information |
900 | what to do if support ended | |
901 | ||
902 | * pam_systemd: on interactive logins, maybe show SUPPORT_END information at | |
3ff1721c | 903 | login time, à la motd |
81a96518 | 904 | |
e1b45a75 | 905 | * sd-boot: instead of unconditionally deriving the ESP to search boot loader |
9c18b363 | 906 | spec entries in from the paths of sd-boot binary, let's optionally allow it |
7227dd81 | 907 | to be configured on sd-boot cmdline + efi var. Use case: embed sd-boot in the |
9c18b363 | 908 | UEFI firmware (for example, ovmf supports that via qemu cmdline option), and |
e1b45a75 | 909 | use it to load stuff from the ESP. |
9c18b363 | 910 | |
b467422b LP |
911 | * mount /var/ from initrd, so that we can apply sysext and stuff before the |
912 | initrd transition. Specifically: | |
913 | 1. There should be a var= kernel cmdline option, matching root= and usr= | |
914 | 2. systemd-gpt-auto-generator should auto-mount /var if it finds it on disk | |
915 | 3. mount.x-initrd mount option in fstab should be implied for /var | |
916 | ||
d486b26f | 917 | * make persistent restarts easier by adding a new setting OpenPersistentFile= |
47b86590 | 918 | or so, which allows opening one or more files that is "persistent" across |
d486b26f LP |
919 | service restarts, hot reboot, cold reboots (depending on configuration): the |
920 | files are created empty on first invocation, and on subsequent invocations | |
921 | the files are reboot. The files would be backed by tmpfs, pmem or /var | |
922 | depending on desired level of persistency. | |
923 | ||
924 | * sd-event: add ability to "chain" event sources. Specifically, add a call | |
925 | sd_event_source_chain(x, y), which will automatically enable event source y | |
1c904337 | 926 | in oneshot mode once x is triggered. Use case: in src/core/mount.c implement |
d486b26f LP |
927 | the /proc/self/mountinfo rescan on SIGCHLD with this: whenever a SIGCHLD is |
928 | seen, trigger the rescan defer event source automatically, and allow it to be | |
929 | dispatched *before* the SIGCHLD is handled (based on priorities). Benefit: | |
930 | dispatch order is strictly controlled by priorities again. (next step: chain | |
931 | event sources to the ratelimit being over) | |
932 | ||
933 | * if we fork of a service with StandardOutput=journal, and it forks off a | |
934 | subprocess that quickly dies, we might not be able to identify the cgroup it | |
935 | comes from, but we can still derive that from the stdin socket its output | |
936 | came from. We apparently don't do that right now. | |
937 | ||
d486b26f LP |
938 | * add ability to set hostname with suffix derived from machine id at boot |
939 | ||
d486b26f LP |
940 | * add PR_SET_DUMPABLE service setting |
941 | ||
027301b4 LP |
942 | * homed/userdb: maybe define a "companion" dir for home directories where apps |
943 | can safely put privileged stuff in. Would not be writable by the user, but | |
944 | still conceptually belong to the user. Would be included in user's quota if | |
7227dd81 | 945 | possible, even if files are not owned by UID of user. Use case: container |
027301b4 LP |
946 | images that owned by arbitrary UIDs, and are owned/managed by the users, but |
947 | are not directly belonging to the user's UID. Goal: we shouldn't place more | |
948 | privileged dirs inside of unprivileged dirs, and thus containers really | |
949 | should not be placed inside of traditional UNIX home dirs (which are owned by | |
3881fd40 | 950 | users themselves) but somewhere else, that is separate, but still close |
027301b4 LP |
951 | by. Inform user code about path to this companion dir via env var, so that |
952 | container managers find it. the ~/.identity file is also a candidate for a | |
953 | file to move there, since it is managed by privileged code (i.e. homed) and | |
954 | not unprivileged code. | |
955 | ||
956 | * given that /etc/ssh/ssh_config.d/ is a thing now, ship a drop-in for that | |
02c914ef | 957 | that hooks up userdbctl ssh-key stuff. |
027301b4 | 958 | |
3a466def LP |
959 | * maybe add support for binding and connecting AF_UNIX sockets in the file |
960 | system outside of the 108ch limit. When connecting, open O_PATH fd to socket | |
961 | inode first, then connect to /proc/self/fd/XYZ. When binding, create symlink | |
962 | to target dir in /tmp, and bind through it. | |
963 | ||
fd74ed23 LP |
964 | * add a proper concept of a "developer" mode, i.e. where cryptographic |
965 | protections of the root OS are weakened after interactive confirmation, to | |
966 | allow hackers to allow their own stuff. idea: allow entering developer mode | |
967 | only via explicit choice in boot menu: i.e. add explicit boot menu item for | |
30fd9a2d | 968 | it. When developer mode is entered, generate a key pair in the TPM2, and add |
fd74ed23 LP |
969 | the public part of it automatically to keychain of valid code signature keys |
970 | on subsequent boots. Then provide a tool to sign code with the key in the | |
30fd9a2d DL |
971 | TPM2. Ensure that boot menu item is the only way to enter developer mode, by |
972 | binding it to locality/PCRs so that keys cannot be generated otherwise. | |
fd74ed23 LP |
973 | |
974 | * services: add support for cryptographically unlocking per-service directories | |
975 | via TPM2. Specifically, for StateDirectory= (and related dirs) use fscrypt to | |
976 | set up the directory so that it can only be accessed if host and app are in | |
977 | order. | |
978 | ||
1d5f14ef LP |
979 | * TPM2: extend unlock policy to protect against version downgrades in signed |
980 | policies: policy probably must take some nvram based generation counter into | |
981 | account that can only monotonically increase and can be used to invalidate | |
982 | old PCR signatures. Otherwise people could downgrade to old signed PCR sets | |
983 | whenever they want. | |
98045d12 LP |
984 | |
985 | * update HACKING.md to suggest developing systemd with the ideas from: | |
986 | https://0pointer.net/blog/testing-my-system-code-in-usr-without-modifying-usr.html | |
987 | https://0pointer.net/blog/running-an-container-off-the-host-usr.html | |
988 | ||
8b825133 LP |
989 | * sd-event: compat wd reuse in inotify code: keep a set of removed watch |
990 | descriptors, and clear this set piecemeal when we see the IN_IGNORED event | |
991 | for it, or when read() returns EAGAIN or on IN_Q_OVERFLOW. Then, whenever we | |
992 | see an inotify wd event check against this set, and if it is contained ignore | |
8ac6b05b | 993 | the event. (to be fully correct this would have to count the occurrences, in |
8b825133 LP |
994 | case the same wd is reused multiple times before we start processing |
995 | IN_IGNORED again) | |
996 | ||
5b06ad51 | 997 | * for vendor-built signed initrds: |
a4b329e6 | 998 | - kernel-install should be able to install encrypted creds automatically for |
62471289 LP |
999 | machine id, root pw, rootfs uuid, resume partition uuid, and place next to |
1000 | EFI kernel, for sd-stub to pick them up. These creds should be locked to | |
1001 | the TPM, and bind to the right PCR the kernel is measured to. | |
11b957b5 LP |
1002 | - kernel-install should be able to pick up initrd sysexts automatically and |
1003 | place them next to EFI kernel, for sd-stub to pick them up. | |
5b06ad51 | 1004 | - systemd-fstab-generator should look for rootfs device to mount in creds |
62471289 | 1005 | - systemd-resume-generator should look for resume partition uuid in creds |
11b957b5 LP |
1006 | - sd-stub: automatically pick up microcode from ESP (/loader/microcode/*) |
1007 | and synthesize initrd from it, and measure it. Signing is not necessary, as | |
1008 | microcode does that on its own. Pass as first initrd to kernel. | |
5b06ad51 | 1009 | |
6fee7849 LP |
1010 | * Maybe extend the service protocol to support handling of some specific SIGRT |
1011 | signal for setting service log level, that carries the level via the | |
1012 | sigqueue() data parameter. Enable this via unit file setting. | |
2df2bb1f | 1013 | |
03ede612 | 1014 | * sd_notify/vsock: maybe support binding to AF_VSOCK in Type=notify services, |
c0da575a LP |
1015 | then passing $NOTIFY_SOCKET and $NOTIFY_GUESTCID with PID1's cid (typically |
1016 | fixed to "2", i.e. the official host cid) and the expected guest cid, for the | |
3a258d3a | 1017 | two sides of the channel. The latter env var could then be used in an |
c0da575a LP |
1018 | appropriate qemu cmdline. That way qemu payloads could talk sd_notify() |
1019 | directly to host service manager. | |
1020 | ||
2cf120f7 LP |
1021 | * sd-device has an API to create an sd_device object from a device id, but has |
1022 | no api to query the device id | |
1023 | ||
1024 | * sd-device should return the devnum type (i.e. 'b' or 'c') via some API for an | |
1025 | sd_device object, so that data passed into sd_device_new_from_devnum() can | |
1026 | also be queried. | |
1027 | ||
5b89bff5 LP |
1028 | * sd-event: optionally, if per-event source rate limit is hit, downgrade |
1029 | priority, but leave enabled, and once ratelimit window is over, upgrade | |
1030 | priority again. That way we can combat event source starvation without | |
1031 | stopping processing events from one source entirely. | |
1032 | ||
e4f92a62 LP |
1033 | * sd-event: similar to existing inotify support add fanotify support (given |
1034 | that apparently new features in this area are only going to be added to the | |
1035 | latter). | |
5b89bff5 LP |
1036 | |
1037 | * sd-event: add 1st class event source for clock changes | |
1038 | ||
1039 | * sd-event: add 1st class event source for timezone changes | |
1040 | ||
9e83d3e4 LP |
1041 | * support uefi/http boots with sd-boot: instead of looking for dropin files in |
1042 | /loader/entries/ dir, look for a file /loader/entries/SHA256SUMS and use that | |
1043 | as directory manifest. The file would be a standard directory listing as | |
1044 | generated by GNU sha256sums. | |
1045 | ||
47a9f917 LP |
1046 | * sd-boot: maybe add support for embedding the various auxiliary resources we |
1047 | look for right in the sd-boot binary. i.e. take inspiration from sd-stub | |
02c914ef | 1048 | logic: allow combining sd-boot via ukify with kernels to enumerate, .conf |
47a9f917 | 1049 | files, drivers, keys to enroll and so on. Then, add whatever we find that way |
7227dd81 | 1050 | to the menu. Use case: allow building a single PE image you can boot into via |
47a9f917 LP |
1051 | UEFI HTTP boot. |
1052 | ||
1053 | * maybe add a new UEFI stub binary "sd-http". It works similar to sd-stub, but | |
1054 | all it does is download a file from a http server, and execute it, after | |
1055 | optionally checking its hash sum. idea would be: combine this "sd-http" stub | |
5bc9ea07 | 1056 | binary with some minimal info about a URL + hash sum, plus .osrel data, and |
47a9f917 LP |
1057 | drop it into the unified kernel dir in the ESP. And bam you have something |
1058 | that is tiny, feels a lot like a unified kernel, but all it does is chainload | |
1059 | the real kernel. benefit: downloading these stubs would be tiny and quick, | |
1060 | hence cheap for enumeration. | |
1061 | ||
d360eafb LP |
1062 | * sysext: measure all activated sysext into a TPM PCR |
1063 | ||
d360eafb LP |
1064 | * systemd-dissect: show available versions inside of a disk image, i.e. if |
1065 | multiple versions are around of the same resource, show which ones. (in other | |
1066 | words: show partition labels). | |
1067 | ||
d360eafb LP |
1068 | * maybe add a generator that reads /proc/cmdline, looks for |
1069 | systemd.pull-raw-portable=, systemd-pull-raw-sysext= and similar switches | |
5bc9ea07 | 1070 | that take a URL as parameter. It then generates service units for |
7227dd81 | 1071 | systemd-pull calls that download these URLs if not installed yet. Use case: |
d360eafb LP |
1072 | invoke a VM or nspawn container in a way it automatically deploys/runs these |
1073 | images as OS payloads. i.e. have a generic OS image you can point to any | |
1074 | payload you like, which is then downloaded, securely verified and run. | |
1075 | ||
e4f92a62 | 1076 | * deprecate cgroupsv1 further (print log message at boot) |
f1a147f2 LP |
1077 | |
1078 | * systemd-dissect: add --cat switch for dumping files such as /etc/os-release | |
1079 | ||
f5d0f21c LP |
1080 | * per-service sandboxing option: ProtectIds=. If used, will overmount |
1081 | /etc/machine-id and /proc/sys/kernel/random/boot_id with synthetic files, to | |
1082 | make it harder for the service to identify the host. Depending on the user | |
1083 | setting it should be fully randomized at invocation time, or a hash of the | |
1084 | real thing, keyed by the unit name or so. Of course, there are other ways to | |
1085 | get these IDs (e.g. journal) or similar ids (e.g. MAC addresses, DMI ids, CPU | |
1086 | ids), so this knob would only be useful in combination with other lockdown | |
1087 | options. Particularly useful for portable services, and anything else that | |
1088 | uses RootDirectory= or RootImage=. (Might also over-mount | |
1089 | /sys/class/dmi/id/*{uuid,serial} with /dev/null). | |
1090 | ||
45cab6e3 LP |
1091 | * kernel-install: |
1092 | - add --all switch for rerunning kernel-install for all installed kernels | |
45cab6e3 | 1093 | |
a57d72ce LP |
1094 | * doc: prep a document explaining resolved's internal objects, i.e. Query |
1095 | vs. Question vs. Transaction vs. Stream and so on. | |
1096 | ||
a5a316e7 LP |
1097 | * doc: prep a document explaining PID 1's internal logic, i.e. transactions, |
1098 | jobs, units | |
1099 | ||
a57d72ce | 1100 | * bootspec: bring UEFI and userspace enumeration of bootspec entries back into |
95150f3f LP |
1101 | sync, i.e. parse out architecture field in sd-boot (currently only done in |
1102 | userspace) | |
a57d72ce | 1103 | |
594f64f4 LP |
1104 | * automatically ignore threaded cgroups in cg_xyz(). |
1105 | ||
e6f48be8 LP |
1106 | * add linker script that implicitly adds symbol for build ID and new coredump |
1107 | json package metadata, and use that when logging | |
1108 | ||
d1666bde | 1109 | * Enable RestrictFileSystems= for all our long-running services (similar: |
f08e143c LP |
1110 | RestrictNetworkInterfaces=) |
1111 | ||
1112 | * Add systemd-analyze security checks for RestrictFileSystems= and | |
1113 | RestrictNetworkInterfaces= | |
1114 | ||
1ccfb792 LP |
1115 | * cryptsetup/homed: implement TOTP authentication backed by TPM2 and its |
1116 | internal clock. | |
1117 | ||
f5ba8115 LP |
1118 | * man: rework os-release(5), and clearly separate our extension-release.d/ and |
1119 | initrd-release parts, i.e. list explicitly which fields are about what. | |
1120 | ||
1121 | * sysext: before applying a sysext, do a superficial validation run so that | |
1122 | things are not rearranged to wildy. I.e. protect against accidental fuckups, | |
1123 | such as masking out /usr/lib/ or so. We should probably refuse if existing | |
1124 | inodes are replaced by other types of inodes or so. | |
1125 | ||
af11e0ef LP |
1126 | * userdb: when synthesizing NSS records, pick "best" password from defined |
1127 | passwords, not just the first. i.e. if there are multiple defined, prefer | |
1128 | unlocked over locked and prefer non-empty over empty. | |
1129 | ||
1130 | * maybe add a tool inspired by the GPT auto discovery spec that runs in the | |
1131 | initrd and rearranges the rootfs hierarchy via bind mounts, if | |
1132 | enabled. Specifically in some top-level dir /@auto/ it will look for | |
1133 | dirs/symlinks/subvolumes that are named after their purpose, and optionally | |
1134 | encode a version as well as assessment counters, and then mount them into the | |
1135 | file system tree to boot into, similar to how we do that for the gpt auto | |
1136 | logic. Maybe then bind mount the original root into /.superior or something | |
1137 | like that (so that update tools can look there). Further discussion in this | |
1138 | thread: | |
1139 | https://lists.freedesktop.org/archives/systemd-devel/2021-November/047059.html | |
1140 | The GPT dissection logic should automatically enable this tool whenever we | |
1141 | detect a specially marked root fs (i.e introduce a new generic root gpt type | |
1142 | for this, that is arch independent). The also implement this in the image | |
1143 | dissection logic, so that nspawn/RootImage= and so on grok it. Maybe make | |
1144 | generic enough so that it can also work for ostrees arrangements. | |
1145 | ||
1146 | * if a path ending in ".auto.d/" is set for RootDirectory=/RootImage= then do a | |
1147 | strverscmp() of everything inside that dir and use that. i.e. implement very | |
1148 | simple version control. Also use this in systemd-nspawn --image= and so on. | |
1149 | ||
1150 | * homed: while a home dir is not activated generate slightly different NSS | |
1151 | records for it, that reports the home dir as "/" and the shell as some binary | |
1152 | provided by us. Then, when an SSH login happens and SSH permits it our binary | |
1153 | is invoked. This binary can then talk to homed and activate the homedir if | |
1154 | it's not around yet, prompting the user for a password. Once that succeeded | |
1155 | we'll switch to the real user record, i.e. home dir and shell, and our tool | |
1156 | exec()s the latter. Net effect: ssh'ing into a homed account will just work: | |
1157 | we'll neatly prompt for the homedir's password if its needed. –– Building on | |
1158 | this we could take this even further: since this tool will potentially have | |
1159 | access to the client's ssh-agent (if ssh-agent forwarding is enabled) we | |
1160 | could implement SSH unlocking of a homedir with that: when enrolling a new | |
1161 | ssh pubkey in a user record we'd ask the ssh-agent to sign some random value | |
1162 | with the privkey, then use that as luks key to unlock the home dir. Will not | |
1163 | work for ECDSA keys since their signatures contain a random component, but | |
1164 | will work for RSA and Ed25519 keys. | |
1165 | ||
bb5464ad LP |
1166 | * add tiny service that decrypts encrypted user records passed via initrd |
1167 | credential logic and drops them into /run where nss-systemd can pick them up, | |
7227dd81 | 1168 | similar to /run/host/userdb/. Use case: drop a root user JSON record there, |
bb5464ad | 1169 | and use it in the initrd to log in as root with locally selected password, |
7227dd81 | 1170 | for debugging purposes. Other use case: boot into qemu with regular user |
11b957b5 | 1171 | mounted from host. maybe put this in systemd-user-sessions.service? |
bb5464ad LP |
1172 | |
1173 | * drop dependency on libcap, replace by direct syscalls based on | |
30fd9a2d | 1174 | CapabilityQuintet we already have. (This likely allows us to drop libcap |
bb5464ad LP |
1175 | dep in the base OS image) |
1176 | ||
bb5464ad LP |
1177 | * add concept for "exitrd" as inverse of "initrd", that we can transition to at |
1178 | shutdown, and has similar security semantics. This should then take the place | |
1179 | of dracut's shutdown logic. Should probably support sysexts too. Care needs | |
1180 | to be taken that the resulting logic ends up in RAM, i.e. is copied out of | |
1181 | on-disk storage. | |
1182 | ||
a07ab1dd LP |
1183 | * userdbd: implement an additional varlink service socket that provides the |
1184 | host user db in restricted form, then allow this to be bind mounted into | |
1185 | sandboxed environments that want the host database in minimal form. All | |
1186 | records would be stripped of all meta info, except the basic UID/name | |
1187 | info. Then use this in portabled environments that do not use PrivateUsers=1. | |
1188 | ||
8825e90a LB |
1189 | * portabled: when extracting unit files and copying to system.attached, if a |
1190 | .p7s is available in the image, use it to protect the system.attached copy | |
1191 | with fs-verity, so that it cannot be tampered with | |
1192 | ||
a5bf435e LP |
1193 | * logind introduce two types of sessions: "heavy" and "light". The former would |
1194 | be our current sessions. But the latter would be a new type of session that | |
1195 | is mostly the same but does not pull in user@.service or wait for it. Then, | |
1196 | allow configuration which type of session is desired via pam_systemd | |
1197 | parameters, and then make user@.service's session one of these "light" ones. | |
1198 | People could then choose to make FTP sessions and suchlike "light" if they | |
1199 | don't want the service manager to be started for that. | |
1200 | ||
da3ab57c LP |
1201 | * /etc/veritytab: allow that the roothash column can be specified as fs path |
1202 | including a path to an AF_UNIX path, similar to how we do things with the | |
1203 | keys of /etc/crypttab. That way people can store/provide the roothash | |
1204 | externally and provide to us on demand only. | |
1205 | ||
3fc0688d | 1206 | * we probably should extend the root verity hash of the root fs into some PCR |
7dad7811 LP |
1207 | on boot. (i.e. maybe add a veritytab option tpm2-measure=12 or so to measure |
1208 | it into PCR 12); Similar: we probably should extend the LUKS volume key of | |
1209 | the root fs into some PCR on boot. (i.e. maybe add a crypttab option | |
1210 | tpm2-measure=15 or so to measure it into PCR 15); once both are in place | |
1211 | update gpt-auto-discovery to generate these by default for the partitions it | |
1212 | discovers. Static vendor stuff should probably end up in PCR 12 (i.e. the | |
1213 | verity hash), with local keys in PCR 15 (i.e. the encryption volume | |
1214 | key). That way, we nicely distinguish resources supplied by the OS vendor | |
1215 | (i.e. sysext, root verity) from those inherently local (i.e. encryption key), | |
1216 | which is useful if they shall be signed separately. | |
3fc0688d | 1217 | |
c0a74f62 LP |
1218 | * in uefi stub: query firmware regarding which PCR banks are being used, store |
1219 | that in EFI var. then use this when enrolling TPM2 in cryptsetup to verify | |
1220 | that the selected PCRs actually are used by firmware. | |
f4529c4d | 1221 | |
322b3b38 LP |
1222 | * rework recursive read-only remount to use new mount API |
1223 | ||
63a185dc | 1224 | * PAM: pick up authentication token from credentials |
9a6549f6 | 1225 | |
9a6549f6 LP |
1226 | * when mounting disk images: if IMAGE_ID/IMAGE_VERSION is set in os-release |
1227 | data in the image, make sure the image filename actually matches this, so | |
1228 | that images cannot be misused. | |
1229 | ||
aca8ecc3 | 1230 | * New udev block device symlink names: |
178d3ff2 | 1231 | /dev/disk/by-parttypelabel/<pttype>-<ptlabel>. Use case: if pt label is used |
aca8ecc3 LP |
1232 | as partition image version string, this is a safe way to reference a specific |
1233 | version of a specific partition type, in particular where related partitions | |
1234 | are processed (e.g. verity + rootfs both named "LennartOS_0.7"). | |
1235 | ||
178d3ff2 LP |
1236 | * sysupdate: |
1237 | - add fuzzing to the pattern parser | |
1238 | - support casync as download mechanism | |
178d3ff2 LP |
1239 | - "systemd-sysupdate update --all" support, that iterates through all components |
1240 | defined on the host, plus all images installed into /var/lib/machines/, | |
1241 | /var/lib/portable/ and so on. | |
1242 | - figure out what to do about system extensions (i.e. they need to imply an | |
1243 | update component, since otherwise system extenion' sysupdate.d/ files would | |
1244 | override the host's update files.) | |
1245 | - Allow invocation with a single transfer definition, i.e. with | |
1246 | --definitions= pointing to a file rather than a dir. | |
1247 | - add ability to disable implicit decompression of downloaded artifacts, | |
1248 | i.e. a Compress=no option in the transfer definitions | |
1249 | ||
cf2ab2e7 LP |
1250 | * in sd-id128: also parse UUIDs in RFC4122 URN syntax (i.e. chop off urn:uuid: prefix) |
1251 | ||
31892e8d LP |
1252 | * DynamicUser= + StateDirectory= → use uid mapping mounts, too, in order to |
1253 | make dirs appear under right UID. | |
1254 | ||
f3e58b55 LP |
1255 | * systemd-sysext: optionally, run it in initrd already, before transitioning |
1256 | into host, to open up possibility for services shipped like that. | |
1257 | ||
24063ba1 | 1258 | * introduce /dev/disk/root/* symlinks that allow referencing partitions on the |
f3e58b55 LP |
1259 | disk the rootfs is on in a reasonably secure way. (or maybe: add |
1260 | /dev/gpt-auto-{home,srv,boot,…} similar in style to /dev/gpt-auto-root as we | |
1261 | already have it. | |
24063ba1 | 1262 | |
7ed72cfa LP |
1263 | * whenever we receive fds via SCM_RIGHTS make sure none got dropped due to the |
1264 | reception limit the kernel silently enforces. | |
1265 | ||
1c904337 | 1266 | * Add service unit setting ConnectStream= which takes IP addresses and connects to them. |
256cfb71 LP |
1267 | |
1268 | * Similar, Load= which takes literal data in text or base64 format, and puts it | |
1269 | into a memfd, and passes that. This enables some fun stuff, such as embedding | |
1270 | bash scripts in unit files, by combining Load= with ExecStart=/bin/bash | |
1271 | /proc/self/fd/3 | |
1272 | ||
66e52d22 LP |
1273 | * add a ConnectSocket= setting to service unit files, that may reference a |
1274 | socket unit, and which will connect to the socket defined therein, and pass | |
1275 | the resulting fd to the service program via socket activation proto. | |
1276 | ||
1277 | * Add a concept of ListenStream=anonymous to socket units: listen on a socket | |
7227dd81 | 1278 | that is deleted in the fs. Use case would be with ConnectSocket= above. |
66e52d22 | 1279 | |
14f7d087 LP |
1280 | * importd: support image signature verification with PKCS#7 + OpenBSD signify |
1281 | logic, as alternative to crummy gpg | |
1282 | ||
9786ba13 LP |
1283 | * add "systemd-analyze debug" + AttachDebugger= in unit files: The former |
1284 | specifies a command to execute; the latter specifies that an already running | |
1285 | "systemd-analyze debug" instance shall be contacted and execution paused | |
1286 | until it gives an OK. That way, tools like gdb or strace can be safely be | |
1287 | invoked on processes forked off PID 1. | |
1288 | ||
38abd1bf LP |
1289 | * expose MS_NOSYMFOLLOW in various places |
1290 | ||
199b097d | 1291 | * credentials system: |
199b097d | 1292 | - acquire from EFI variable? |
30fd9a2d | 1293 | - acquire via ask-password? |
199b097d LP |
1294 | - acquire creds via keyring? |
1295 | - pass creds via keyring? | |
1296 | - pass creds via memfd? | |
1297 | - acquire + decrypt creds from pkcs11? | |
1298 | - make systemd-cryptsetup acquire pw via creds logic | |
1299 | - make PAMName= acquire pw via creds logic | |
1300 | - make macsec/wireguard code in networkd read key via creds logic | |
1301 | - make gatwayd/remote read key via creds logic | |
1302 | - add sd_notify() command for flushing out creds not needed anymore | |
1f066ce2 LB |
1303 | - make user manager instances create and use a user-specific key (the one in |
1304 | /var/lib is root-only) and add --user switch to systemd-creds to use it | |
199b097d | 1305 | |
199b097d LP |
1306 | * add tpm.target or so which is delayed until TPM2 device showed up in case |
1307 | firmware indicates there is one. | |
07eabc2b | 1308 | |
07eabc2b LB |
1309 | * TPM2: auto-reenroll in cryptsetup, as fallback for hosed firmware upgrades |
1310 | and such | |
1311 | ||
1312 | * introduce a new group to own TPM devices | |
80670e74 | 1313 | |
02c914ef | 1314 | * cryptsetup: add option for automatically removing empty password slot on boot |
80670e74 | 1315 | |
7d7c75f1 | 1316 | * cryptsetup: optionally, when run during boot-up and password is never |
80670e74 LP |
1317 | entered, and we are on battery power (or so), power off machine again |
1318 | ||
80670e74 LP |
1319 | * cryptsetup: when waiting for FIDO2/PKCS#11 token, tell plymouth that, and |
1320 | allow plymouth to abort the waiting and enter pw instead | |
7d7c75f1 | 1321 | |
07eabc2b LB |
1322 | * make cryptsetup lower --iter-time |
1323 | ||
1324 | * cryptsetup: allow encoding key directly in /etc/crypttab, maybe with a | |
1325 | "base64:" prefix. Useful in particular for pkcs11 mode. | |
1326 | ||
1327 | * cryptsetup: reimplement the mkswap/mke2fs in cryptsetup-generator to use | |
1328 | systemd-makefs.service instead. | |
1329 | ||
1330 | * cryptsetup: | |
1331 | - cryptsetup-generator: allow specification of passwords in crypttab itself | |
1332 | - support rd.luks.allow-discards= kernel cmdline params in cryptsetup generator | |
1333 | ||
08d33656 | 1334 | * systemd-analyze netif that explains predictable interface (or networkctl) |
d775d8e6 | 1335 | |
816d460a LP |
1336 | * Add service setting to run a service within the specified VRF. i.e. do the |
1337 | equivalent of "ip vrf exec". | |
1338 | ||
f461a28d | 1339 | * special case some calls of chase() to use openat2() internally, so |
e59d030f LP |
1340 | that the kernel does what we otherwise do. |
1341 | ||
f461a28d | 1342 | * add a new flag to chase() that stops chasing once the first missing |
6dbfbc46 LB |
1343 | component is found and then allows the caller to create the rest. |
1344 | ||
fc733bed LP |
1345 | * make use of new glibc 2.32 APIs sigabbrev_np() and strerrorname_np(). |
1346 | ||
effefa30 LP |
1347 | * if /usr/bin/swapoff fails due to OOM, log a friendly explanatory message about it |
1348 | ||
f1eb0ccd LP |
1349 | * pid1: support new clone3() fork-into-cgroup feature |
1350 | ||
7cc8fb3e LP |
1351 | * pid1: also remove PID files of a service when the service starts, not just |
1352 | when it exits | |
1353 | ||
cdfd8537 LP |
1354 | * make us use dynamically fewer deps for containers in general purpose distros: |
1355 | o turn into dlopen() deps: | |
cdfd8537 | 1356 | - kmod-libs (only when called from PID 1) |
a52dc0b6 | 1357 | - libblkid (only in RootImage= handling in PID 1, but not elsewhere) |
cdfd8537 LP |
1358 | - libpam (only when called from PID 1) |
1359 | - bzip2, xz, lz4 (always — gzip and zstd should probably stay static deps the way they are, | |
1360 | since they are so basic and our defaults) | |
1361 | o move into separate libsystemd-shared-iptables.so .so | |
1362 | - iptables-libs (only used by nspawn + networkd) | |
1363 | ||
77169ed0 LP |
1364 | * seccomp: maybe use seccomp_merge() to merge our filters per-arch if we can. |
1365 | Apparently kernel performance is much better with fewer larger seccomp | |
1366 | filters than with more smaller seccomp filters. | |
1367 | ||
bfafec25 LP |
1368 | * systemd-path: add ESP and XBOOTLDR path. Add "private" runtime/state/cache dir enum, |
1369 | mapping to $RUNTIME_DIRECTORY, $STATE_DIRECTORY and such | |
1370 | ||
112bed84 LP |
1371 | * seccomp: by default mask x32 ABI system wide on x86-64. it's on its way out |
1372 | ||
1373 | * seccomp: don't install filters for ABIs that are masked anyway for the | |
1374 | specific service | |
1375 | ||
cbe952fe LP |
1376 | * busctl: maybe expose a verb "ping" for pinging a dbus service to see if it |
1377 | exists and responds. | |
1378 | ||
91fc013f | 1379 | * socket units: allow creating a udev monitor socket with ListenDevices= or so, |
86b52a39 | 1380 | with matches, then activate app through that passing socket over |
91fc013f | 1381 | |
7e8facb3 | 1382 | * unify on openssl: |
070f7370 LB |
1383 | - kill gnutls support in resolved |
1384 | - figure out what to do about libmicrohttpd, which has a hard dependency on | |
1385 | gnutls | |
1386 | - port fsprg over to a dlopen lib, then switch it to openssl | |
492f91d8 | 1387 | |
492f91d8 LP |
1388 | * add growvol and makevol options for /etc/crypttab, similar to |
1389 | x-systemd.growfs and x-systemd-makefs. | |
1390 | ||
77b19caf LP |
1391 | * userdb: allow username prefix searches in varlink API, allow realname and |
1392 | realname substr searches in varlink API | |
006c44c1 | 1393 | |
76410e98 LP |
1394 | * userdb: allow uid/gid range checks |
1395 | ||
2a4be3c5 | 1396 | * userdb: allow existence checks |
006c44c1 | 1397 | |
f1eb0ccd | 1398 | * pid1: activation by journal search expression |
006c44c1 | 1399 | |
492f91d8 LP |
1400 | * when switching root from initrd to host, set the machine_id env var so that |
1401 | if the host has no machine ID set yet we continue to use the random one the | |
1402 | initrd had set. | |
1403 | ||
173c7873 | 1404 | * sd-event: add native support for P_ALL waitid() watching, then move PID 1 to |
84a1ff94 | 1405 | it for reaping assigned but unknown children. This needs to some special care |
173c7873 LP |
1406 | to operate somewhat sensibly in light of priorities: P_ALL will return |
1407 | arbitrary processes, regardless of the priority we want to watch them with, | |
1408 | hence on each event loop iteration check all processes which we shall watch | |
1409 | with higher prio explicitly, and then watch the entire rest with P_ALL. | |
1410 | ||
1411 | * tweak sd-event's child watching: keep a prioq of children to watch and use | |
1412 | waitid() only on the children with the highest priority until one is waitable | |
1413 | and ignore all lower-prio ones from that point on | |
1414 | ||
173c7873 LP |
1415 | * maybe introduce xattrs that can be set on the root dir of the root fs |
1416 | partition that declare the volatility mode to use the image in. Previously I | |
1417 | thought marking this via GPT partition flags but that's not ideal since | |
1418 | that's outside of the LUKS encryption/verity verification, and we probably | |
1419 | shouldn't operate in a volatile mode unless we got told so from a trusted | |
1420 | source. | |
1421 | ||
97e5cc88 LP |
1422 | * coredump: maybe when coredumping read a new xattr from /proc/$PID/exe that |
1423 | may be used to mark a whole binary as non-coredumpable. Would fix: | |
1424 | https://bugs.freedesktop.org/show_bug.cgi?id=69447 | |
1425 | ||
f3e361c1 LP |
1426 | * teach parse_timestamp() timezones like the calendar spec already knows it |
1427 | ||
c846b233 LP |
1428 | * beef up s2h to implement a battery watch loop: instead of entering |
1429 | hibernation unconditionally after coming back from resume make a decision | |
1430 | based on the battery load level: if battery level is above a specific | |
1431 | threshold, go to suspend again, only hibernate if below it. This means we'd | |
1432 | stick to suspend usually, but fall back to hibernation only when battery runs | |
1433 | empty (well, subject to our sampling interval). Related to this, check if we | |
1434 | can make ACPI _BTP (i.e. /sys/class/power_supply/*/alarm) work for us too, | |
1435 | i.e. see if it can wake up machines from suspend, so that we could resume | |
1436 | automatically when the system is low on power and move automatically to | |
1437 | hibernation mode. (see | |
1438 | https://uefi.org/sites/default/files/resources/ACPI%206_2_A_Sept29.pdf | |
1439 | section 10.2.2.8 and | |
1440 | https://docs.microsoft.com/en-us/windows-hardware/design/device-experiences/modern-standby-wake-sources | |
1441 | at the end). | |
1442 | ||
6fe23ff3 LB |
1443 | * We should probably replace /etc/rc.d/README with a symlink to doc |
1444 | content. After all it is constant vendor data. | |
9c230b8f | 1445 | |
c6526b8d | 1446 | * maybe add kernel cmdline params: to force random seed crediting |
312dc153 | 1447 | |
ac5dca64 LP |
1448 | * introduce a new per-process uuid, similar to the boot id, the machine id, the |
1449 | invocation id, that is derived from process creds, specifically a hashed | |
1450 | combination of AT_RANDOM + getpid() + the starttime from | |
1451 | /proc/self/status. Then add these ids implicitly when logging. Deriving this | |
1452 | uuid from these three things has the benefit that it can be derived easily | |
1453 | from /proc/$PID/ in a stable, and unique way that changes on both fork() and | |
1454 | exec(). | |
1455 | ||
1456 | * let's not GC a unit while its ratelimits are still pending | |
1457 | ||
5daeeecf LP |
1458 | * when killing due to service watchdog timeout maybe detect whether target |
1459 | process is under ptracing and then log loudly and continue instead. | |
1460 | ||
5802d977 LP |
1461 | * make rfkill uaccess controllable by default, i.e. steal rule from |
1462 | gnome-bluetooth and friends | |
1463 | ||
03ccc7f0 LP |
1464 | * make MAINPID= message reception checks even stricter: if service uses User=, |
1465 | then check sending UID and ignore message if it doesn't match the user or | |
1466 | root. | |
1467 | ||
1468 | * maybe trigger a uevent "change" on a device if "systemctl reload xyz.device" | |
1469 | is issued. | |
1470 | ||
db3cea22 LP |
1471 | * when importing an fs tree with machined, optionally apply userns-rec-chown |
1472 | ||
1473 | * when importing an fs tree with machined, complain if image is not an OS | |
1474 | ||
707b3fbd LP |
1475 | * Maybe introduce a helper safe_exec() or so, which is to execve() which |
1476 | safe_fork() is to fork(). And then make revert the RLIMIT_NOFILE soft limit | |
1477 | to 1K implicitly, unless explicitly opted-out. | |
d7b659ef | 1478 | |
d238709c | 1479 | * rework seccomp/nnp logic that even if User= is used in combination with |
d96c081a LP |
1480 | a seccomp option we don't have to set NNP. For that, change uid first whil |
1481 | keeping CAP_SYS_ADMIN, then apply seccomp, the drop cap. | |
1482 | ||
8a7cf157 LP |
1483 | * when no locale is configured, default to UEFI's PlatformLang variable |
1484 | ||
d96c081a LP |
1485 | * add a new syscall group "@esoteric" for more esoteric stuff such as bpf() and |
1486 | usefaultd() and make systemd-analyze check for it. | |
1487 | ||
06898123 | 1488 | * paranoia: whenever we process passwords, call mlock() on the memory |
309c6b19 | 1489 | first. i.e. look for all places we use free_and_erasep() and |
9ae4ef49 | 1490 | augment them with mlock(). Also use MADV_DONTDUMP. |
8b213bf1 | 1491 | Alternatively (preferably?) use memfd_secret(). |
06898123 | 1492 | |
32875617 LP |
1493 | * Move RestrictAddressFamily= to the new cgroup create socket |
1494 | ||
06898123 LP |
1495 | * optionally: turn on cgroup delegation for per-session scope units |
1496 | ||
42e18088 LP |
1497 | * sd-boot: optionally, show boot menu when previous default boot item has |
1498 | non-zero "tries done" count | |
1499 | ||
8f2eb730 LP |
1500 | * augment CODE_FILE=, CODE_LINE= with something like CODE_BASE= or so which |
1501 | contains some identifier for the project, which allows us to include | |
1502 | clickable links to source files generating these log messages. The identifier | |
1503 | could be some abberviated URL prefix or so (taking inspiration from Go | |
1504 | imports). For example, for systemd we could use | |
1505 | CODE_BASE=github.com/systemd/systemd/blob/98b0b1123cc or so which is | |
1506 | sufficient to build a link by prefixing "http://" and suffixing the | |
1507 | CODE_FILE. | |
1508 | ||
8f2eb730 LP |
1509 | * Augment MESSAGE_ID with MESSAGE_BASE, in a similar fashion so that we can |
1510 | make clickable links from log messages carrying a MESSAGE_ID, that lead to | |
1511 | some explanatory text online. | |
1512 | ||
7bd4bcf7 LP |
1513 | * maybe extend .path units to expose fanotify() per-mount change events |
1514 | ||
d1797c42 | 1515 | * hibernate/s2h: if swap is on weird storage and refuse if so |
c633b0a6 | 1516 | |
070d0ac9 LP |
1517 | * cgroups: use inotify to get notified when somebody else modifies cgroups |
1518 | owned by us, then log a friendly warning. | |
1519 | ||
e44924f5 LP |
1520 | * beef up log.c with support for stripping ANSI sequences from strings, so that |
1521 | it is OK to include them in log strings. This would be particularly useful so | |
1522 | that our log messages could contain clickable links for example for unit | |
1523 | files and suchlike we operate on. | |
070d0ac9 | 1524 | |
07eabc2b LB |
1525 | * importd: add ability download images for portabled + sysext |
1526 | ||
8c4c2dfc LP |
1527 | * add support for "portablectl attach http://foobar.com/waaa.raw (i.e. importd integration) |
1528 | ||
8c4c2dfc LP |
1529 | * sync dynamic uids/gids between host+portable srvice (i.e. if DynamicUser=1 is set for a service, make sure that the |
1530 | selected user is resolvable in the service even if it ships its own /etc/passwd) | |
1531 | ||
5da19043 | 1532 | * Fix DECIMAL_STR_MAX or DECIMAL_STR_WIDTH. One includes a trailing NUL, the |
63a185dc | 1533 | other doesn't. What a disaster. Probably to exclude it. |
5da19043 | 1534 | |
16270697 LP |
1535 | * Check that users of inotify's IN_DELETE_SELF flag are using it properly, as |
1536 | usually IN_ATTRIB is the right way to watch deleted files, as the former only | |
1537 | fires when a file is actually removed from disk, i.e. the link count drops to | |
1538 | zero and is not open anymore, while the latter happens when a file is | |
1539 | unlinked from any dir. | |
1540 | ||
bd1b3f75 | 1541 | * port systemctl, busctl, … over to format-table.[ch]'s table formatters |
5da19043 | 1542 | |
7bc756ff LP |
1543 | * pid1: lock image configured with RootDirectory=/RootImage= using the usual nspawn semantics while the unit is up |
1544 | ||
1545 | * add --vacuum-xyz options to coredumpctl, matching those journalctl already has. | |
1546 | ||
53c70a27 | 1547 | * add CopyFile= or so as unit file setting that may be used to copy files or |
5238e957 | 1548 | directory trees from the host to the services RootImage= and RootDirectory= |
53c70a27 LP |
1549 | environment. Which we can use for /etc/machine-id and in particular |
1550 | /etc/resolv.conf. Should be smart and do something useful on read-only | |
2aed63f4 | 1551 | images, for example fall back to read-only bind mounting the file instead. |
53c70a27 | 1552 | |
d9b50610 LP |
1553 | * show invocation ID in systemd-run output |
1554 | ||
1555 | * bypass SIGTERM state in unit files if KillSignal is SIGKILL | |
1556 | ||
9711b1ad LP |
1557 | * add proper dbus APIs for the various sd_notify() commands, such as MAINPID=1 |
1558 | and so on, which would mean we could report errors and such. | |
1559 | ||
6b7b0f39 LP |
1560 | * introduce DefaultSlice= or so in system.conf that allows changing where we |
1561 | place our units by default, i.e. change system.slice to something | |
1562 | else. Similar, ManagerSlice= should exist so that PID1's own scope unit could | |
1563 | be moved somewhere else too. Finally machined and logind should get similar | |
1564 | options so that it is possible to move user session scopes and machines to a | |
7227dd81 | 1565 | different slice too by default. Use case: people who want to put resources on |
6b7b0f39 LP |
1566 | the entire system, with the exception of one specific service. See: |
1567 | https://lists.freedesktop.org/archives/systemd-devel/2018-February/040369.html | |
1568 | ||
aa79f932 LP |
1569 | * maybe rework get_user_creds() to query the user database if $SHELL is used |
1570 | for root, but only then. | |
586a8e93 | 1571 | |
46099c9e LP |
1572 | * calenderspec: add support for week numbers and day numbers within a |
1573 | year. This would allow us to define "bi-weekly" triggers safely. | |
1574 | ||
c3cd7cc9 LP |
1575 | * sd-bus: add vtable flag, that may be used to request client creds implicitly |
1576 | and asynchronously before dispatching the operation | |
1577 | ||
d0e5db44 ZJS |
1578 | * sd-bus: parse addresses given in sd_bus_set_addresses immediately and not |
1579 | only when used. Add unit tests. | |
1580 | ||
06345858 LP |
1581 | * make use of ethtool veth peer info in machined, for automatically finding out |
1582 | host-side interface pointing to the container. | |
1583 | ||
48f1b5e5 LP |
1584 | * add some special mode to LogsDirectory=/StateDirectory=… that allows |
1585 | declaring these directories without necessarily pulling in deps for them, or | |
1586 | creating them when starting up. That way, we could declare that | |
1587 | systemd-journald writes to /var/log/journal, which could be useful when we | |
1588 | doing disk usage calculations and so on. | |
1589 | ||
899feb72 | 1590 | * deprecate RootDirectoryStartOnly= in favour of a new ExecStart= prefix char |
c6009ff0 | 1591 | |
5f7ecd61 | 1592 | * support projid-based quota in machinectl for containers |
5962e9db | 1593 | |
17b6f896 LP |
1594 | * add a way to lock down cgroup migration: a boolean, which when set for a unit |
1595 | makes sure the processes in it can never migrate out of it | |
1596 | ||
17b6f896 LP |
1597 | * blog about fd store and restartable services |
1598 | ||
1599 | * document Environment=SYSTEMD_LOG_LEVEL=debug drop-in in debugging document | |
1600 | ||
370f9c21 LP |
1601 | * rework ExecOutput and ExecInput enums so that EXEC_OUTPUT_NULL loses its |
1602 | magic meaning and is no longer upgraded to something else if set explicitly. | |
1603 | ||
17b6f896 LP |
1604 | * in the long run: permit a system with /etc/machine-id linked to /dev/null, to |
1605 | make it lose its identity, i.e. be anonymous. For this we'd have to patch | |
1606 | through the whole tree to make all code deal with the case where no machine | |
1607 | ID is available. | |
1608 | ||
1609 | * optionally, collect cgroup resource data, and store it in per-unit RRD files, | |
1610 | suitable for processing with rrdtool. Add bus API to access this data, and | |
1611 | possibly implement a CPULoad property based on it. | |
1612 | ||
e7e4a258 LP |
1613 | * beef up pam_systemd to take unit file settings such as cgroups properties as |
1614 | parameters | |
1615 | ||
63a185dc | 1616 | * maybe hook up xfs/ext4 quotactl() with services? i.e. automatically manage |
d51c4fca | 1617 | the quota of the user indicated in User= via unit file settings, like the |
5962e9db LP |
1618 | other resource management concepts. Would mix nicely with DynamicUser=1. Or |
1619 | alternatively, do this with projids, so that we can also cover services | |
1620 | running as root. Quota should probably cover all the special dirs such as | |
1621 | StateDirectory=, LogsDirectory=, CacheDirectory=, as well as RootDirectory= if it | |
1622 | is set, plus the whole disk space any image configured with RootImage=. | |
6fc373ee | 1623 | |
9a92a97a LP |
1624 | * In DynamicUser= mode: before selecting a UID, use disk quota APIs on relevant |
1625 | disks to see if the UID is already in use. | |
1626 | ||
d73b607d LP |
1627 | * Add AddUser= setting to unit files, similar to DynamicUser=1 which however |
1628 | creates a static, persistent user rather than a dynamic, transient user. We | |
1629 | can leverage code from sysusers.d for this. | |
1630 | ||
fd63f36c LP |
1631 | * add some optional flag to ReadWritePaths= and friends, that has the effect |
1632 | that we create the dir in question when the service is started. Example: | |
1633 | ||
1634 | ReadWritePaths=:/var/lib/foobar | |
1635 | ||
f59d1da8 LP |
1636 | * Add ExecMonitor= setting. May be used multiple times. Forks off a process in |
1637 | the service cgroup, which is supposed to monitor the service, and when it | |
1638 | exits the service is considered failed by its monitor. | |
1639 | ||
33bac67b LP |
1640 | * track the per-service PAM process properly (i.e. as an additional control |
1641 | process), so that it may be queried on the bus and everything. | |
1642 | ||
1643 | * add a new "debug" job mode, that is propagated to unit_start() and for | |
1644 | services results in two things: we raise SIGSTOP right before invoking | |
1645 | execve() and turn off watchdog support. Then, use that to implement | |
1646 | "systemd-gdb" for attaching to the start-up of any system service in its | |
1647 | natural habitat. | |
1648 | ||
d1666bde LP |
1649 | * gpt-auto logic: support encrypted swap, add kernel cmdline option to force |
1650 | it, and honour a gpt bit about it, plus maybe a configuration file | |
8eb7383b | 1651 | |
b5bdbcd5 LP |
1652 | * add a percentage syntax for TimeoutStopSec=, e.g. TimeoutStopSec=150%, and |
1653 | then use that for the setting used in user@.service. It should be understood | |
1654 | relative to the configured default value. | |
1655 | ||
d21494ea LP |
1656 | * enable LockMLOCK to take a percentage value relative to physical memory |
1657 | ||
04397464 | 1658 | * Permit masking specific netlink APIs with RestrictAddressFamily= |
d82047be | 1659 | |
d82047be LP |
1660 | * define gpt header bits to select volatility mode |
1661 | ||
42d61ded LP |
1662 | * ProtectClock= (drops CAP_SYS_TIMES, adds seecomp filters for settimeofday, adjtimex), sets DeviceAllow o /dev/rtc |
1663 | ||
04397464 | 1664 | * ProtectTracing= (drops CAP_SYS_PTRACE, blocks ptrace syscall, makes /sys/kernel/tracing go away) |
42d61ded | 1665 | |
04397464 | 1666 | * ProtectMount= (drop mount/umount/pivot_root from seccomp, disallow fuse via DeviceAllow, imply Mountflags=slave) |
563a69f4 | 1667 | |
04397464 | 1668 | * ProtectKeyRing= to take keyring calls away |
2c5f2958 | 1669 | |
8ce9b83a | 1670 | * RemoveKeyRing= to remove all keyring entries of the specified user |
a46eac1b LP |
1671 | |
1672 | * ProtectReboot= that masks reboot() and kexec_load() syscalls, prohibits kill | |
1673 | on PID 1 with the relevant signals, and makes relevant files in /sys and | |
1674 | /proc (such as the sysrq stuff) unavailable | |
1675 | ||
88511a37 LB |
1676 | * Support ReadWritePaths/ReadOnlyPaths/InaccessiblePaths in systemd --user instances |
1677 | via the new unprivileged Landlock LSM (https://landlock.io) | |
1678 | ||
e40a326c | 1679 | * make sure the ratelimit object can deal with USEC_INFINITY as way to turn off things |
89f193fa | 1680 | |
8ce9b83a LP |
1681 | * in nss-systemd, if we run inside of RootDirectory= with PrivateUsers= set, |
1682 | find a way to map the User=/Group= of the service to the right name. This way | |
1683 | a user/group for a service only has to exist on the host for the right | |
1684 | mapping to work. | |
1685 | ||
2c5f2958 LP |
1686 | * add bus API for creating unit files in /etc, reusing the code for transient units |
1687 | ||
1688 | * add bus API to remove unit files from /etc | |
1689 | ||
1690 | * add bus API to retrieve current unit file contents (i.e. implement "systemctl cat" on the bus only) | |
1691 | ||
b8c7afdf LP |
1692 | * rework fopen_temporary() to make use of open_tmpfile_linkable() (problem: the |
1693 | kernel doesn't support linkat() that replaces existing files, currently) | |
1694 | ||
1e555cb5 LP |
1695 | * transient units: don't bother with actually setting unit properties, we |
1696 | reload the unit file anyway | |
1697 | ||
c8048350 LP |
1698 | * optionally, also require WATCHDOG=1 notifications during service start-up and shutdown |
1699 | ||
3d39e6e5 LP |
1700 | * cache sd_event_now() result from before the first iteration... |
1701 | ||
bd098bce LP |
1702 | * PID1: find a way how we can reload unit file configuration for |
1703 | specific units only, without reloading the whole of systemd | |
1704 | ||
f9bf1b8f | 1705 | * add an explicit parser for LimitRTPRIO= that verifies |
de7399eb | 1706 | the specified range and generates sane error messages for incorrect |
f9bf1b8f | 1707 | specifications. |
de7399eb | 1708 | |
3efc8c72 LP |
1709 | * when we detect that there are waiting jobs but no running jobs, do something |
1710 | ||
e6a26d8c LP |
1711 | * PID 1 should send out sd_notify("WATCHDOG=1") messages (for usage in the --user mode, and when run via nspawn) |
1712 | ||
a2088fd0 | 1713 | * there's probably something wrong with having user mounts below /sys, |
5238e957 | 1714 | as we have for debugfs. for example, src/core/mount.c handles mounts |
a2088fd0 | 1715 | prefixed with /sys generally special. |
41d6f3bf | 1716 | https://lists.freedesktop.org/archives/systemd-devel/2015-June/032962.html |
a2088fd0 | 1717 | |
8aa20381 LP |
1718 | * fstab-generator: default to tmpfs-as-root if only usr= is specified on the kernel cmdline |
1719 | ||
931bc195 | 1720 | * docs: bring https://www.freedesktop.org/wiki/Software/systemd/MyServiceCantGetRealtime up to date |
b18d23d7 | 1721 | |
60d17b74 LP |
1722 | * add a job mode that will fail if a transaction would mean stopping |
1723 | running units. Use this in timedated to manage the NTP service | |
1724 | state. | |
41d6f3bf | 1725 | https://lists.freedesktop.org/archives/systemd-devel/2015-April/030229.html |
60d17b74 | 1726 | |
477e75ef LP |
1727 | * The udev blkid built-in should expose a property that reflects |
1728 | whether media was sensed in USB CF/SD card readers. This should then | |
1729 | be used to control SYSTEMD_READY=1/0 so that USB card readers aren't | |
1730 | picked up by systemd unless they contain a medium. This would mirror | |
1731 | the behaviour we already have for CD drives. | |
1732 | ||
c3a0d00d LP |
1733 | * hostnamectl: show root image uuid |
1734 | ||
d2f81fb0 | 1735 | * Find a solution for SMACK capabilities stuff: |
41d6f3bf | 1736 | https://lists.freedesktop.org/archives/systemd-devel/2014-December/026188.html |
98cd2651 | 1737 | |
0a86c1a9 | 1738 | * synchronize console access with BSD locks: |
41d6f3bf | 1739 | https://lists.freedesktop.org/archives/systemd-devel/2014-October/024582.html |
0a86c1a9 | 1740 | |
e031c227 | 1741 | * as soon as we have sender timestamps, revisit coalescing multiple parallel daemon reloads: |
41d6f3bf | 1742 | https://lists.freedesktop.org/archives/systemd-devel/2014-December/025862.html |
0a86c1a9 | 1743 | |
8514b677 LP |
1744 | * figure out when we can use the coarse timers |
1745 | ||
8483d73f LP |
1746 | * maybe allow timer units with an empty Units= setting, so that they |
1747 | can be used for resuming the system but nothing else. | |
1748 | ||
25e773ee | 1749 | * what to do about udev db binary stability for apps? (raw access is not an option) |
b857e042 | 1750 | |
720652b3 | 1751 | * exponential backoff in timesyncd when we cannot reach a server |
42aeb14a | 1752 | |
720652b3 | 1753 | * timesyncd: add ugly bus calls to set NTP servers per-interface, for usage by NM |
e25b5a8d | 1754 | |
6bd7941e TG |
1755 | * add systemd.abort_on_kill or some other such flag to send SIGABRT instead of SIGKILL |
1756 | (throughout the codebase, not only PID1) | |
1757 | ||
07eabc2b LB |
1758 | * drop nss-myhostname in favour of nss-resolve? |
1759 | ||
9d6db739 | 1760 | * resolved: |
9d6db739 | 1761 | - mDNS/DNS-SD |
ccc3e8a1 LP |
1762 | - service registration |
1763 | - service/domain/types browsing | |
0f47ed0a | 1764 | - avahi compat |
9d6db739 | 1765 | - DNS-SD service registration from socket units |
e25b5a8d DH |
1766 | - resolved should optionally register additional per-interface LLMNR |
1767 | names, so that for the container case we can establish the same name | |
1768 | (maybe "host") for referencing the server, everywhere. | |
720652b3 | 1769 | - allow clients to request DNSSEC for a single lookup even if DNSSEC is off (?) |
3efb871a | 1770 | - hook up resolved with machined-based address resolution |
3f77a1b1 | 1771 | |
e25b5a8d | 1772 | * refcounting in sd-resolve is borked |
e2a69298 | 1773 | |
a940778f LP |
1774 | * add new gpt type for btrfs volumes |
1775 | ||
3de03738 LP |
1776 | * generator that automatically discovers btrfs subvolumes, identifies their purpose based on some xattr on them. |
1777 | ||
37efac5d LP |
1778 | * a way for container managers to turn off getty starting via $container_headless= or so... |
1779 | ||
7348b3ad LP |
1780 | * figure out a nice way how we can let the admin know what child/sibling unit causes cgroup membership for a specific unit |
1781 | ||
81429136 KS |
1782 | * For timer units: add some mechanisms so that timer units that trigger immediately on boot do not have the services |
1783 | they run added to the initial transaction and thus confuse Type=idle. | |
e107ed18 | 1784 | |
edb2935c LP |
1785 | * add bus api to query unit file's X fields. |
1786 | ||
6a3f892a | 1787 | * gpt-auto-generator: |
2a781fc9 | 1788 | - Define new partition type for encrypted swap? Support probed LUKS for encrypted swap? |
6a3f892a | 1789 | - Make /home automount rather than mount? |
6a3f892a | 1790 | |
65026403 LP |
1791 | * add generator that pulls in systemd-network from containers when |
1792 | CAP_NET_ADMIN is set, more than the loopback device is defined, even | |
1793 | when it is otherwise off | |
f8901862 | 1794 | |
f9bf1b8f | 1795 | * MessageQueueMessageSize= (and suchlike) should use parse_iec_size(). |
eda8f067 | 1796 | |
af1082b0 LP |
1797 | * implement Distribute= in socket units to allow running multiple |
1798 | service instances processing the listening socket, and open this up | |
1799 | for ReusePort= | |
1800 | ||
f38afcd0 | 1801 | * cgroups: |
f38afcd0 | 1802 | - implement per-slice CPUFairScheduling=1 switch |
f38afcd0 LP |
1803 | - introduce high-level settings for RT budget, swappiness |
1804 | - how to reset dynamically changed unit cgroup attributes sanely? | |
1805 | - when reloading configuration, apply new cgroup configuration | |
1806 | - when recursively showing the cgroup hierarchy, optionally also show | |
1807 | the hierarchies of child processes | |
97046f2f LP |
1808 | - add settings for cgroup.max.descendants and cgroup.max.depth, |
1809 | maybe use them for user@.service | |
0bee65f0 | 1810 | |
f38afcd0 | 1811 | * transient units: |
f38afcd0 | 1812 | - add field to transient units that indicate whether systemd or somebody else saves/restores its settings, for integration with libvirt |
ebcf1f97 | 1813 | |
718db961 LP |
1814 | * when we detect low battery and no AC on boot, show pretty splash and refuse boot |
1815 | ||
718db961 LP |
1816 | * libsystemd-journal, libsystemd-login, libudev: add calls to easily attach these objects to sd-event event loops |
1817 | ||
966204e0 LP |
1818 | * be more careful what we export on the bus as (usec_t) 0 and (usec_t) -1 |
1819 | ||
41644622 LP |
1820 | * 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 |
1821 | ||
19aadacf JE |
1822 | * If we try to find a unit via a dangling symlink, generate a clean |
1823 | error. Currently, we just ignore it and read the unit from the search | |
df5f6971 LP |
1824 | path anyway. |
1825 | ||
04397464 | 1826 | * refuse boot if /usr/lib/os-release is missing or /etc/machine-id cannot be set up |
fcba531e | 1827 | |
bdeeb6b5 LP |
1828 | * man: the documentation of Restart= currently is very misleading and suggests the tools from ExecStartPre= might get restarted. |
1829 | ||
1830 | * load .d/*.conf dropins for device units | |
1831 | ||
07eabc2b | 1832 | * There's currently no way to cancel fsck (used to be possible via C-c or c on the console) |
f38afcd0 | 1833 | |
07eabc2b | 1834 | * add option to sockets to avoid activation. Instead just drop packets/connections, see http://cyberelk.net/tim/2012/02/15/portreserve-systemd-solution/ |
eb01ba5d | 1835 | |
07eabc2b LB |
1836 | * make sure systemd-ask-password-wall does not shutdown systemd-ask-password-console too early |
1837 | ||
1838 | * verify that the AF_UNIX sockets of a service in the fs still exist | |
1839 | when we start a service in order to avoid confusion when a user | |
1840 | assumes starting a service is enough to make it accessible | |
1841 | ||
1842 | * Make it possible to set the keymap independently from the font on | |
1843 | the kernel cmdline. Right now setting one resets also the other. | |
1844 | ||
1845 | * and a dbus call to generate target from current state | |
1846 | ||
1847 | * investigate whether the gnome pty helper should be moved into systemd, to provide cgroup support. | |
1848 | ||
1849 | * dot output for --test showing the 'initial transaction' | |
1850 | ||
1851 | * be able to specify a forced restart of service A where service B depends on, in case B | |
1852 | needs to be auto-respawned? | |
1853 | ||
1854 | * pid1: | |
1855 | - When logging about multiple units (stopping BoundTo units, conflicts, etc.), | |
1856 | log both units as UNIT=, so that journalctl -u triggers on both. | |
1857 | - generate better errors when people try to set transient properties | |
1858 | that are not supported... | |
41d6f3bf | 1859 | https://lists.freedesktop.org/archives/systemd-devel/2015-February/028076.html |
7227dd81 | 1860 | - maybe introduce WantsMountsFor=? Use case: |
41d6f3bf | 1861 | https://lists.freedesktop.org/archives/systemd-devel/2015-January/027729.html |
07eabc2b LB |
1862 | - recreate systemd's D-Bus private socket file on SIGUSR2 |
1863 | - move PAM code into its own binary | |
1864 | - when we automatically restart a service, ensure we restart its rdeps, too. | |
1865 | - hide PAM options in fragment parser when compile time disabled | |
1866 | - Support --test based on current system state | |
1867 | - 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(). | |
1868 | - after deserializing sockets in socket.c we should reapply sockopts and things | |
1869 | - drop PID 1 reloading, only do reexecing (difficult: Reload() | |
1870 | currently is properly synchronous, Reexec() is weird, because we | |
1871 | cannot delay the response properly until we are back, so instead of | |
1872 | being properly synchronous we just keep open the fd and close it | |
1873 | when done. That means clients do not get a successful method reply, | |
1874 | but much rather a disconnect on success. | |
1875 | - when breaking cycles drop sysv services first, then services from /run, then from /etc, then from /usr | |
1876 | - when a bus name of a service disappears from the bus make sure to queue further activation requests | |
1877 | - maybe introduce CoreScheduling=yes/no to optionally set a PR_SCHED_CORE cookie, so that all | |
1878 | processes in a service's cgroup share the same cookie and are guaranteed not to share SMT cores | |
1879 | with other units https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/admin-guide/hw-vuln/core-scheduling.rst | |
1880 | ||
1881 | * unit files: | |
1882 | - allow port=0 in .socket units | |
1883 | - maybe introduce ExecRestartPre= | |
07eabc2b LB |
1884 | - implement Register= switch in .socket units to enable registration |
1885 | in Avahi, RPC and other socket registration services. | |
1886 | - allow Type=simple with PIDFile= | |
1887 | https://bugzilla.redhat.com/show_bug.cgi?id=723942 | |
1888 | - allow writing multiple conditions in unit files on one line | |
1889 | - introduce Type=pid-file | |
1890 | - add a concept of RemainAfterExit= to scope units | |
1891 | - Allow multiple ExecStart= for all Type= settings, so that we can cover rescue.service nicely | |
1892 | - add verification of [Install] section to systemd-analyze verify | |
1893 | ||
1894 | * timer units: | |
63a185dc | 1895 | - timer units should get the ability to trigger when DST changes |
07eabc2b LB |
1896 | - Modulate timer frequency based on battery state |
1897 | ||
1898 | * add libsystemd-password or so to query passwords during boot using the password agent logic | |
1899 | ||
1900 | * clean up date formatting and parsing so that all absolute/relative timestamps we format can also be parsed | |
1901 | ||
1902 | * on shutdown: move utmp, wall, audit logic all into PID 1 (or logind?), get rid of systemd-update-utmp-runlevel | |
1903 | ||
1904 | * make repeated alt-ctrl-del presses printing a dump | |
1905 | ||
07eabc2b | 1906 | * currently x-systemd.timeout is lost in the initrd, since crypttab is copied into dracut, but fstab is not |
461bd8e4 | 1907 | |
ab8e074c LP |
1908 | * add a pam module that passes the hdd passphrase into the PAM stack and then expires it, for usage by gdm auto-login. |
1909 | ||
1910 | * add a pam module that on password changes updates any LUKS slot where the password matches | |
1911 | ||
fff87a35 | 1912 | * test/: |
20d52ab6 | 1913 | - add unit tests for config_parse_device_allow() |
b8b4d3dd | 1914 | |
b5c03638 | 1915 | * seems that when we follow symlinks to units we prefer the symlink |
d28315e4 | 1916 | destination path over /etc and /usr. We should not do that. Instead |
b5c03638 LP |
1917 | /etc should always override /run+/usr and also any symlink |
1918 | destination. | |
1919 | ||
eece8c6f LP |
1920 | * when isolating, try to figure out a way how we implicitly can order |
1921 | all units we stop before the isolating unit... | |
1922 | ||
356ce991 LP |
1923 | * teach ConditionKernelCommandLine= globs or regexes (in order to match foobar={no,0,off}) |
1924 | ||
6daebf9e ZJS |
1925 | * Add ConditionDirectoryNotEmpty= handle non-absoute paths as a search path or add |
1926 | ConditionConfigSearchPathNotEmpty= or different syntax? See the discussion starting at | |
1927 | https://github.com/systemd/systemd/pull/15109#issuecomment-607740136. | |
1928 | ||
1ccfb792 LP |
1929 | * BootLoaderSpec: Define a way how an installer can figure out whether a BLS |
1930 | compliant boot loader is installed. | |
795607b2 | 1931 | |
7227dd81 | 1932 | * think about requeuing jobs when daemon-reload is issued? use case: |
0be8342c LP |
1933 | the initrd issues a reload after fstab from the host is accessible |
1934 | and we might want to requeue the mounts local-fs acquired through | |
1935 | that automatically. | |
1936 | ||
e5ec62c5 | 1937 | * systemd-inhibit: make taking delay locks useful: support sending SIGINT or SIGTERM on PrepareForSleep() |
54c31a79 | 1938 | |
ccddd104 | 1939 | * remove any syslog support from log.c — we probably cannot do this before split-off udev is gone for good |
826872b6 | 1940 | |
3679d112 LP |
1941 | * shutdown logging: store to EFI var, and store to USB stick? |
1942 | ||
356ce991 | 1943 | * merge unit_kill_common() and unit_kill_context() |
490b7e47 | 1944 | |
07eabc2b LB |
1945 | * add a dependency on standard-conf.xml and other included files to man pages |
1946 | ||
1947 | * MountFlags=shared acts as MountFlags=slave right now. | |
1948 | ||
1949 | * properly handle loop back mounts via fstab, especially regards to fsck/passno | |
1950 | ||
1951 | * initialize the hostname from the fs label of /, if /etc/hostname does not exist? | |
1952 | ||
1953 | * sd-bus: | |
1954 | - EBADSLT handling | |
1955 | - GetAllProperties() on a non-existing object does not result in a failure currently | |
1956 | - port to sd-resolve for connecting to TCP dbus servers | |
1957 | - 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 | |
1958 | - see if we can drop more message validation on the sending side | |
1959 | - add API to clone sd_bus_message objects | |
1960 | - longer term: priority inheritance | |
1961 | - dbus spec updates: | |
1962 | - NameLost/NameAcquired obsolete | |
07eabc2b LB |
1963 | - path escaping |
1964 | - update systemd.special(7) to mention that dbus.socket is only about the compatibility socket now | |
1965 | ||
1966 | * sd-event | |
1967 | - allow multiple signal handlers per signal? | |
1968 | - document chaining of signal handler for SIGCHLD and child handlers | |
1969 | - define more intervals where we will shift wakeup intervals around in, 1h, 6h, 24h, ... | |
1970 | - maybe support iouring as backend, so that we allow hooking read and write | |
1971 | operations instead of IO ready events into event loops. See considerations | |
1972 | here: | |
1973 | http://blog.vmsplice.net/2020/07/rethinking-event-loop-integration-for.html | |
1974 | ||
1975 | * dbus: when a unit failed to load (i.e. is in UNIT_ERROR state), we | |
1976 | should be able to safely try another attempt when the bus call LoadUnit() is invoked. | |
1977 | ||
9140404a ZJS |
1978 | * document org.freedesktop.MemoryAllocation1 |
1979 | ||
07eabc2b LB |
1980 | * maybe do not install getty@tty1.service symlink in /etc but in /usr? |
1981 | ||
1982 | * print a nicer explanation if people use variable/specifier expansion in ExecStart= for the first word | |
1983 | ||
1984 | * mount: turn dependency information from /proc/self/mountinfo into dependency information between systemd units. | |
1985 | ||
b44be3ec | 1986 | * EFI: |
b44be3ec LP |
1987 | - honor language efi variables for default language selection (if there are any?) |
1988 | - honor timezone efi variables for default timezone selection (if there are any?) | |
466784c8 | 1989 | - change bootctl to be backed by systemd-bootd to control temporary and persistent default boot goal plus efi variables |
631427d6 | 1990 | * bootctl |
631427d6 | 1991 | - recognize the case when not booted on EFI |
e4181484 | 1992 | |
07eabc2b | 1993 | * bootctl,sd-boot: actually honour the "architecture" key |
e9fd44b7 | 1994 | |
07eabc2b | 1995 | * bootctl: |
483091b0 | 1996 | - show whether UEFI audit mode is available |
07eabc2b LB |
1997 | - teach it to prepare an ESP wholesale, i.e. with mkfs.vfat invocation |
1998 | - teach it to copy in unified kernel images and maybe type #1 boot loader spec entries from host | |
f620a368 | 1999 | |
b44be3ec LP |
2000 | * logind: |
2001 | - logind: optionally, ignore idle-hint logic for autosuspend, block suspend as long as a session is around | |
b44be3ec LP |
2002 | - logind: wakelock/opportunistic suspend support |
2003 | - Add pretty name for seats in logind | |
2004 | - logind: allow showing logout dialog from system? | |
f38afcd0 LP |
2005 | - add Suspend() bus calls which take timestamps to fix double suspend issues when somebody hits suspend and closes laptop quickly. |
2006 | - if pam_systemd is invoked by su from a process that is outside of a | |
2007 | any session we should probably just become a NOP, since that's | |
2008 | usually not a real user session but just some system code that just | |
2009 | needs setuid(). | |
279f0366 LP |
2010 | - logind: make the Suspend()/Hibernate() bus calls wait for the for |
2011 | the job to be completed. before returning, so that clients can wait | |
2012 | for "systemctl suspend" to finish to know when the suspending is | |
2013 | complete. | |
2014 | - logind: when the power button is pressed short, just popup a | |
2015 | logout dialog. If it is pressed for 1s, do the usual | |
2016 | shutdown. Inspiration are Macs here. | |
28423d9a | 2017 | - expose "Locked" property on logind session objects |
e25b5a8d DH |
2018 | - maybe allow configuration of the StopTimeout for session scopes |
2019 | - rename session scope so that it includes the UID. THat way | |
2020 | the session scope can be arranged freely in slices and we don't have | |
2021 | make assumptions about their slice anymore. | |
2022 | - follow PropertiesChanged state more closely, to deal with quick logouts and | |
2023 | relogins | |
77b19caf | 2024 | - (optionally?) spawn seat-manager@$SEAT.service whenever a seat shows up that as CanGraphical set |
4a5f779f ZJS |
2025 | - expose details of boot entries on the bus. In particular, it should be possible |
2026 | to query the list of boot entry titles that bootctl / sd-boot would show. | |
2027 | Currently we only expose their identifiers. | |
07eabc2b LB |
2028 | |
2029 | * move multiseat vid/pid matches from logind udev rule to hwdb | |
2030 | ||
2031 | * logind: rework pam_logind to also do a bus call in case of invocation from | |
2032 | user@.service, which returns the XDG_RUNTIME_DIR value, and make this | |
2033 | behaviour selectable via pam module option. | |
2034 | ||
2035 | * delay activation of logind until somebody logs in, or when /dev/tty0 pulls it | |
2036 | in or lingering is on (so that containers don't bother with it until PAM is used). also exit-on-idle | |
2037 | ||
b44be3ec | 2038 | * journal: |
57f2a947 | 2039 | - consider introducing implicit _TTY= + _PPID= + _EUID= + _EGID= + _FSUID= + _FSGID= fields |
b44be3ec LP |
2040 | - journald: also get thread ID from client, plus thread name |
2041 | - journal: when waiting for journal additions in the client always sleep at least 1s or so, in order to minimize wakeups | |
2042 | - add API to close/reopen/get fd for journal client fd in libsystemd-journal. | |
2aed63f4 | 2043 | - fall back to /dev/log based logging in libsystemd-journal, if we cannot log natively? |
b44be3ec | 2044 | - declare the local journal protocol stable in the wiki interface chart |
b44be3ec LP |
2045 | - sd-journal: speed up sd_journal_get_data() with transparent hash table in bg |
2046 | - journald: when dropping msgs due to ratelimit make sure to write | |
2047 | "dropped %u messages" not only when we are about to print the next | |
5238e957 | 2048 | message that works, but already after a short timeout |
b44be3ec LP |
2049 | - check if we can make journalctl by default use --follow mode inside of less if called without args? |
2050 | - maybe add API to send pairs of iovecs via sd_journal_send | |
f47ec8eb | 2051 | - journal: add a setgid "systemd-journal" utility to invoke from libsystemd-journal, which passes fds via STDOUT and does PK access |
02c914ef | 2052 | - journalctl: support negative filtering, i.e. FOOBAR!="waldo", |
b44be3ec | 2053 | and !FOOBAR for events without FOOBAR. |
06847d0f | 2054 | - journal: store timestamp of journal_file_set_offline() in the header, |
038cf334 | 2055 | so it is possible to display when the file was last synced. |
b44be3ec LP |
2056 | - 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. |
2057 | - journal: find a way to allow dropping history early, based on priority, other rules | |
2058 | - journal: When used on NFS, check payload hashes | |
b44be3ec LP |
2059 | - journald: add kernel cmdline option to disable ratelimiting for debug purposes |
2060 | - refuse taking lower-case variable names in sd_journal_send() and friends. | |
2061 | - journald: we currently rotate only after MaxUse+MaxFilesize has been reached. | |
2062 | - journal: deal nicely with byte-by-byte copied files, especially regards header | |
b44be3ec | 2063 | - journal: sanely deal with entries which are larger than the individual file size, but where the components would fit |
601d9d6f | 2064 | - Replace utmp, wtmp, btmp, and lastlog completely with journal |
f38afcd0 | 2065 | - journalctl: instead --after-cursor= maybe have a --cursor=XYZ+1 syntax? |
f38afcd0 LP |
2066 | - when a kernel driver logs in a tight loop, we should ratelimit that too. |
2067 | - journald: optionally, log debug messages to /run but everything else to /var | |
2068 | - journald: when we drop syslog messages because the syslog socket is | |
2069 | full, make sure to write how many messages are lost as first thing | |
2070 | to syslog when it works again. | |
279f0366 LP |
2071 | - journald: allow per-priority and per-service retention times when rotating/vacuuming |
2072 | - journald: make use of uid-range.h to managed uid ranges to split | |
2073 | journals in. | |
2074 | - journalctl: add the ability to look for the most recent process of a binary. journalctl /usr/bin/X11 --pid=-1 or so... | |
2075 | - improve journalctl performance by loading journal files | |
2076 | lazily. Encode just enough information in the file name, so that we | |
2077 | do not have to open it to know that it is not interesting for us, for | |
2078 | the most common operations. | |
e25b5a8d | 2079 | - man: document that corrupted journal files is nothing to act on |
e25b5a8d DH |
2080 | - rework journald sigbus stuff to use mutex |
2081 | - Set RLIMIT_NPROC for systemd-journal-xyz, and all other of our | |
2082 | services that run under their own user ids, and use User= (but only | |
2083 | in a world where userns is ubiquitous since otherwise we cannot | |
2084 | invoke those daemons on the host AND in a container anymore). Also, | |
2085 | if LimitNPROC= is used without User= we should warn and refuse | |
2086 | operation. | |
2087 | - journalctl --verify: don't show files that are currently being | |
2088 | written to as FAIL, but instead show that their are being written to. | |
2089 | - add journalctl -H that talks via ssh to a remote peer and passes through | |
2090 | binary logs data | |
e25b5a8d | 2091 | - add a version of --merge which also merges /var/log/journal/remote |
e25b5a8d DH |
2092 | - journalctl: -m should access container journals directly by enumerating |
2093 | them via machined, and also watch containers coming and going. | |
2094 | Benefit: nspawn --ephemeral would start working nicely with the journal. | |
2095 | - assign MESSAGE_ID to log messages about failed services | |
06847d0f | 2096 | - check if loop in decompress_blob_xz() is necessary |
b44be3ec | 2097 | |
07eabc2b LB |
2098 | * journald: support RFC3164 fully for the incoming syslog transport, see |
2099 | https://github.com/systemd/systemd/issues/19251#issuecomment-816601955 | |
2100 | ||
2101 | * Hook up journald's FSS logic with TPM2: seal the verification disk by | |
2102 | time-based policy, so that the verification key can remain on host and ve | |
2103 | validated via TPM. | |
2104 | ||
0134afe3 ZJS |
2105 | * rework journalctl -M to be based on a machined method that generates a mount |
2106 | fd of the relevant journal dirs in the container with uidmapping applied to | |
2107 | allow the host to read it, while making everything read-only. | |
2108 | ||
2109 | * journald: add varlink service that allows subscribing to certain log events, | |
2110 | for example matching by message ID, or log level returns a list of journal | |
2111 | cursors as they happen. | |
2112 | ||
2113 | * journald: also collect CLOCK_BOOTTIME timestamps per log entry. Then, derive | |
2114 | "corrected" CLOCK_REALTIME information on display from that and the timestamp | |
2115 | info of the newest entry of the specific boot (as identified by the boot | |
2116 | ID). This way, if a system comes up without a valid clock but acquires a | |
2117 | better clock later, we can "fix" older entry timestamps on display, by | |
2118 | calculating backwards. We cannot use CLOCK_MONOTONIC for this, since it does | |
2119 | not account for suspend phases. This would then also enable us to correct the | |
2120 | kmsg timestamping we consume (where we erroneously assume the clock was in | |
2121 | CLOCK_MONOTONIC, but it actually is CLOCK_BOOTTIME as per kernel). | |
2122 | ||
2123 | * in journald, write out a recognizable log record whenever the system clock is | |
2124 | changed ("stepped"), and in timesyncd whenever we acquire an NTP fix | |
2125 | ("slewing"). Then, in journalctl for each boot time we come across, find | |
2126 | these records, and use the structured info they include to display | |
cbcf76b1 | 2127 | "corrected" wallclock time, as calculated from the monotonic timestamp in the |
0134afe3 ZJS |
2128 | log record, adjusted by the delta declared in the structured log record. |
2129 | ||
2130 | * in journald: whenever we start a new journal file because the boot ID | |
2131 | changed, let's generate a recognizable log record containing info about old | |
2132 | and new ID. Then, when displaying log stream in journalctl look for these | |
2133 | records, to be able to order them. | |
2134 | ||
2135 | * journald: generate recognizable log events whenever we shutdown journald | |
2136 | cleanly, and when we migrate run → var. This way tools can verify that a | |
2137 | previous boot terminated cleanly, because either of these two messages must | |
2138 | be safely written to disk, then. | |
2139 | ||
2140 | * hook up journald with TPMs? measure new journal records to the TPM in regular | |
2141 | intervals, validate the journal against current TPM state with that. (taking | |
2142 | inspiration from IMA log) | |
2143 | ||
2144 | * sd-journal puts a limit on parallel journal files to view at once. journald | |
2145 | should probably honour that same limit (JOURNAL_FILES_MAX) when vacuuming to | |
2146 | ensure we never generate more files than we can actually view. | |
2147 | ||
2148 | * maybe add a tool that displays most recent journal logs as QR code to scan | |
2149 | off screen and run it automatically on boot failures, emergency logs and | |
2150 | such. Use DRM APIs directly, see | |
2151 | https://github.com/dvdhrm/docs/blob/master/drm-howto/modeset.c for an example | |
2152 | for doing that. | |
2153 | ||
2154 | * maybe implicitly attach monotonic+realtime timestamps to outgoing messages in | |
2155 | log.c and sd-journal-send | |
2156 | ||
2157 | * journalctl/timesyncd: whenever timesyncd acquires a synchronization from NTP, | |
2158 | create a structured log entry that contains boot ID, monotonic clock and | |
2159 | realtime clock (I mean, this requires no special work, as these three fields | |
2160 | are implicit). Then in journalctl when attempting to display the realtime | |
2161 | timestamp of a log entry, first search for the closest later log entry | |
2162 | of this kinda that has a matching boot id, and convert the monotonic clock | |
2163 | timestamp of the entry to the realtime clock using this info. This way we can | |
2164 | retroactively correct the wallclock timestamps, in particular for systems | |
2165 | without RTC, i.e. where initially wallclock timestamps carry rubbish, until | |
2166 | an NTP sync is acquired. | |
2167 | ||
2168 | * introduce per-unit (i.e. per-slice, per-service) journal log size limits. | |
07eabc2b LB |
2169 | |
2170 | * journald: do journal file writing out-of-process, with one writer process per | |
2171 | client UID, so that synthetic hash table collisions can slow down a specific | |
2172 | user's journal stream down but not the others. | |
2173 | ||
2174 | * tweak journald context caching. In addition to caching per-process attributes | |
2175 | keyed by PID, cache per-cgroup attributes (i.e. the various xattrs we read) | |
2176 | keyed by cgroup path, and guarded by ctime changes. This should provide us | |
2177 | with a nice speed-up on services that have many processes running in the same | |
2178 | cgroup. | |
2179 | ||
2180 | * maybe add call sd_journal_set_block_timeout() or so to set SO_SNDTIMEO for | |
2181 | the sd-journal logging socket, and, if the timeout is set to 0, sets | |
2182 | O_NONBLOCK on it. That way people can control if and when to block for | |
2183 | logging. | |
2184 | ||
2185 | * journalctl: make sure -f ends when the container indicated by -M terminates | |
2186 | ||
2187 | * journald: sigbus API via a signal-handler safe function that people may call | |
2188 | from the SIGBUS handler | |
2189 | ||
590171d1 ZJS |
2190 | * add a test if all entries in the catalog are properly formatted. |
2191 | (Adding dashes in a catalog entry currently results in the catalog entry | |
2192 | being silently skipped. journalctl --update-catalog must warn about this, | |
2193 | and we should also have a unit test to check that all our message are OK.) | |
2194 | ||
0134afe3 ZJS |
2195 | * build short web pages out of each catalog entry, build them along with man |
2196 | pages, and include hyperlinks to them in the journal output | |
2197 | ||
07eabc2b LB |
2198 | * homed: |
2199 | - when user tries to log into record signed by unrecognized key, automatically add key to our chain after polkit auth | |
2200 | - rollback when resize fails mid-operation | |
2201 | - GNOME's side for forget key on suspend (requires rework so that lock screen runs outside of uid) | |
07eabc2b LB |
2202 | - update LUKS password on login if we find there's a password that unlocks the JSON record but not the LUKS device. |
2203 | - create on activate? | |
2204 | - properties: icon url?, preferred session type?, administrator bool (which translates to 'wheel' membership)?, address?, telephone?, vcard?, samba stuff?, parental controls? | |
2205 | - communicate clearly when usb stick is safe to remove. probably involves | |
2206 | beefing up logind to make pam session close hook synchronous and wait until | |
2207 | systemd --user is shut down. | |
2208 | - logind: maybe keep a "busy fd" as long as there's a non-released session around or the user@.service | |
2209 | - maybe make automatic, read-only, time-based reflink-copies of LUKS disk | |
2210 | images (and btrfs snapshots of subvolumes) (think: time machine) | |
2211 | - distinguish destroy / remove (i.e. currently we can unregister a user, unregister+remove their home directory, but not just remove their home directory) | |
2212 | - in systemd's PAMName= logic: query passwords with ssh-askpassword, so that we can make "loginctl set-linger" mode work | |
2213 | - fingerprint authentication, pattern authentication, … | |
2214 | - make sure "classic" user records can also be managed by homed | |
2215 | - make size of $XDG_RUNTIME_DIR configurable in user record | |
2216 | - query password from kernel keyring first | |
2217 | - update even if record is "absent" | |
07eabc2b LB |
2218 | - move acct mgmt stuff from pam_systemd_home to pam_systemd? |
2219 | - when "homectl --pkcs11-token-uri=" is used, synthesize ssh-authorized-keys records for all keys we have private keys on the stick for | |
2220 | - make slice for users configurable (requires logind rework) | |
2221 | - logind: populate auto-login list bus property from PKCS#11 token | |
2222 | - when determining state of a LUKS home directory, check DM suspended sysfs file | |
a11e7c0b LB |
2223 | - when homed is in use, maybe start the user session manager in a mount namespace with MS_SLAVE, |
2224 | so that mounts propagate down but not up - eg, user A setting up a backup volume | |
2225 | doesn't mean user B sees it | |
9c53de8b | 2226 | - use credentials logic/TPM2 logic to store homed signing key |
9c53de8b LP |
2227 | - permit multiple user record signing keys to be used locally, and pick |
2228 | the right one for signing records automatically depending on a pre-existing | |
2229 | signature | |
2230 | - add a way to "adopt" a home directory, i.e. strip foreign signatures | |
2231 | and insert a local signature instead. | |
2232 | - as an extension to the directory+subvolume backend: if located on | |
2233 | especially marked fs, then sync down password into LUKS header of that fs, | |
2234 | and always verify passwords against it too. Bootstrapping is a problem | |
2235 | though: if no one is logged in (or no other user even exists yet), how do you | |
2236 | unlock the volume in order to create the first user and add the first pw. | |
2237 | - support new FS_IOC_ADD_ENCRYPTION_KEY ioctl for setting up fscrypt | |
2238 | - maybe pre-create ~/.cache as subvol so that it can have separate quota | |
2239 | easily? | |
9c53de8b LP |
2240 | - add a switch to homectl (maybe called --first-boot) where it will check if |
2241 | any non-system users exist, and if not prompts interactively for basic user | |
2242 | info, mimicking systemd-firstboot. Then, place this in a service that runs | |
2243 | after systemd-homed, but before gdm and friends, as a simple, barebones | |
2244 | fallback logic to get a regular user created on uninitialized systems. | |
2245 | - store PKCS#11 + FIDO2 token info in LUKS2 header, compatible with | |
2246 | systemd-cryptsetup, so that it can unlock homed volumes | |
6d975fe7 LP |
2247 | - maybe make all *.home files owned by `systemd-home` user or so, so that we |
2248 | can easily set overall quota for all users | |
2249 | - on login, if we can't fallocate initially, but rebalance is on, then allow | |
2250 | login in discard mode, then immediately rebalance, then turn off discard | |
f1a147f2 | 2251 | - extend user records with optional "bulk" data. Specifically, a user |
288bd406 | 2252 | avatar/photo or so. This data should be stored along with the user record, |
f1a147f2 LP |
2253 | but probably shouldn't be part of the record itself, since it might be |
2254 | large. | |
07eabc2b | 2255 | |
07eabc2b LB |
2256 | * add a new switch --auto-definitions=yes/no or so to systemd-repart. If |
2257 | specified, synthesize a definition automatically if we can: enlarge last | |
2258 | partition on disk, but only if it is marked for growing and not read-only. | |
2259 | ||
2df2bb1f | 2260 | * systemd-repart: read LUKS encryption key from $CREDENTIALS_DIRECTORY |
07eabc2b LB |
2261 | |
2262 | * systemd-repart: add a switch to factory reset the partition table without | |
2263 | immediately applying the new configuration again. i.e. --factory-reset=leave | |
2264 | or so. (this is useful to factory reset an image, then putting it into | |
2265 | another machine, ensuring that luks key is generated on new machine, not old) | |
2266 | ||
2267 | * systemd-repart: support setting up dm-integrity with HMAC | |
2268 | ||
2269 | * systemd-repart: maybe remove half-initialized image on failure. It fails | |
2270 | if the output file exists, so a repeated invocation will usually fail if | |
2271 | something goes wrong on the way. | |
2272 | ||
07eabc2b LB |
2273 | * systemd-repart: by default generate minimized partition tables (i.e. tables |
2274 | that only cover the space actually used, excluding any free space at the | |
2275 | end), in order to maximize dd'ability. Requires libfdisk work, see | |
2276 | https://github.com/karelzak/util-linux/issues/907 | |
2277 | ||
2278 | * systemd-repart: MBR partition table support. Care needs to be taken regarding | |
2279 | Type=, so that partition definitions can sanely apply to both the GPT and the | |
2280 | MBR case. Idea: accept syntax "Type=gpt:home mbr:0x83" for setting the types | |
2281 | for the two partition types explicitly. And provide an internal mapping so | |
2282 | that "Type=linux-generic" maps to the right types for both partition tables | |
2283 | automatically. | |
2284 | ||
2285 | * systemd-repart: allow sizing partitions as factor of available RAM, so that | |
2286 | we can reasonably size swap partitions for hibernation. | |
2287 | ||
2288 | * systemd-repart: allow boolean option that ensures that if existing partition | |
2289 | doesn't exist within the configured size bounds the whole command fails. This | |
2290 | is useful to implement ESP vs. XBOOTLDR schemes in installers: have one set | |
2291 | of repart files for the case where ESP is large enough and one where it isn't | |
2292 | and XBOOTLDR is added in instead. Then apply the former first, and if it | |
2293 | fails to apply use the latter. | |
2294 | ||
2295 | * systemd-repart: add per-partition option to never reuse existing partition | |
2296 | and always create anew even if matching partition already exists. | |
2297 | ||
2298 | * systemd-repart: add per-partition option to fail if partition already exist, | |
2299 | i.e. is not added new. Similar, add option to fail if partition does not exist yet. | |
2300 | ||
2301 | * systemd-repart: allow disabling growing of specific partitions, or making | |
2302 | them (think ESP: we don't ever want to grow it, since we cannot resize vfat) | |
d01d9197 | 2303 | Also add option to disable operation via kernel command line. |
07eabc2b LB |
2304 | |
2305 | * systemd-repart: make it a static checker during early boot for existence and | |
2306 | absence of other partitions for trusted boot environments | |
2307 | ||
92e72028 | 2308 | * systemd-repart: add support for SD_GPT_FLAG_GROWFS also on real systems, i.e. |
d01d9197 ZJS |
2309 | generate some unit to actually enlarge the fs after growing the partition |
2310 | during boot. | |
2311 | ||
2312 | * systemd-repart: do not print "Successfully resized …" when no change was done. | |
2313 | ||
b44be3ec | 2314 | * document: |
8b8f2591 | 2315 | - document that deps in [Unit] sections ignore Alias= fields in |
b44be3ec LP |
2316 | [Install] units of other units, unless those units are disabled |
2317 | - man: clarify that time-sync.target is not only sysv compat but also useful otherwise. Same for similar targets | |
b44be3ec | 2318 | - document that service reload may be implemented as service reexec |
f38afcd0 LP |
2319 | - add a man page containing packaging guidelines and recommending usage of things like Documentation=, PrivateTmp=, PrivateNetwork= and ReadOnlyDirectories=/etc /usr. |
2320 | - document systemd-journal-flush.service properly | |
f38afcd0 LP |
2321 | - documentation: recommend to connect the timer units of a service to the service via Also= in [Install] |
2322 | - man: document the very specific env the shutdown drop-in tools live in | |
5cf821ac ZJS |
2323 | - man: add more examples to man pages, |
2324 | - in particular an example how to do the equivalent of switching runlevels | |
f38afcd0 | 2325 | - man: maybe sort directives in man pages, and take sections from --help and apply them to man too |
17ec531f | 2326 | - document root=gpt-auto properly |
b44be3ec LP |
2327 | |
2328 | * systemctl: | |
b44be3ec LP |
2329 | - add systemctl switch to dump transaction without executing it |
2330 | - Add a verbose mode to "systemctl start" and friends that explains what is being done or not done | |
2331 | - "systemctl disable" on a static unit prints no message and does | |
2332 | nothing. "systemctl enable" does nothing, and gives a bad message | |
2333 | about it. Should fix both to print nice actionable messages. | |
2334 | - print nice message from systemctl --failed if there are no entries shown, and hook that into ExecStartPre of rescue.service/emergency.service | |
2335 | - add new command to systemctl: "systemctl system-reexec" which reexecs as many daemons as virtually possible | |
d28315e4 | 2336 | - systemctl enable: fail if target to alias into does not exist? maybe show how many units are enabled afterwards? |
b44be3ec | 2337 | - systemctl: "Journal has been rotated since unit was started." message is misleading |
61233823 | 2338 | - systemctl status output should include list of triggering units and their status |
f38afcd0 | 2339 | |
07eabc2b LB |
2340 | * introduce an option (or replacement) for "systemctl show" that outputs all |
2341 | properties as JSON, similar to busctl's new JSON output. In contrast to that | |
2342 | it should skip the variant type string though. | |
8b04b925 | 2343 | |
07eabc2b LB |
2344 | * Add a "systemctl list-units --by-slice" mode or so, which rearranges the |
2345 | output of "systemctl list-units" slightly by showing the tree structure of | |
2346 | the slices, and the units attached to them. | |
a19554ed | 2347 | |
07eabc2b LB |
2348 | * add "systemctl wait" or so, which does what "systemd-run --wait" does, but |
2349 | for all units. It should be both a way to pin units into memory as well as a | |
2350 | wait to retrieve their exit data. | |
a7a3f28b | 2351 | |
07eabc2b LB |
2352 | * show whether a service has out-of-date configuration in "systemctl status" by |
2353 | using mtime data of ConfigurationDirectory=. | |
08f95888 | 2354 | |
07eabc2b LB |
2355 | * "systemctl preset-all" should probably order the unit files it |
2356 | operates on lexicographically before starting to work, in order to | |
2357 | ensure deterministic behaviour if two unit files conflict (like DMs | |
2358 | do, for example) | |
dcfc4b2e | 2359 | |
07eabc2b LB |
2360 | * add "systemctl start -v foobar.service" that shows logs of a service |
2361 | while the start command runs. This is non-trivial to do without | |
2362 | races though, since we should flush out all journal messages before | |
2363 | returning from the "systemctl stop". | |
71ef24d0 | 2364 | |
07eabc2b LB |
2365 | * systemctl: if some operation fails, show log output? |
2366 | ||
2367 | * Add a new verb "systemctl top" | |
2368 | ||
2369 | * unit install: | |
2370 | - "systemctl mask" should find all names by which a unit is accessible | |
2371 | (i.e. by scanning for symlinks to it) and link them all to /dev/null | |
1b89884b | 2372 | |
b44be3ec | 2373 | * nspawn: |
e25b5a8d DH |
2374 | - emulate /dev/kmsg using CUSE and turn off the syslog syscall |
2375 | with seccomp. That should provide us with a useful log buffer that | |
2376 | systemd can log to during early boot, and disconnect container logs | |
2377 | from the kernel's logs. | |
2378 | - as soon as networkd has a bus interface, hook up --network-interface=, | |
2379 | --network-bridge= with networkd, to trigger netdev creation should an | |
2380 | interface be missing | |
e25b5a8d DH |
2381 | - a nice way to boot up without machine id set, so that it is set at boot |
2382 | automatically for supporting --ephemeral. Maybe hash the host machine id | |
2383 | together with the machine name to generate the machine id for the container | |
2384 | - fix logic always print a final newline on output. | |
2385 | https://github.com/systemd/systemd/pull/272#issuecomment-113153176 | |
2386 | - should optionally support receiving WATCHDOG=1 messages from its payload | |
2387 | PID 1... | |
e25b5a8d DH |
2388 | - optionally automatically add FORWARD rules to iptables whenever nspawn is |
2389 | running, remove them when shut down. | |
71b77f06 CB |
2390 | - add support for sysext extensions, too. i.e. a new --extension= switch that |
2391 | takes one or more arguments, and applies the extensions already during | |
2392 | startup. | |
2393 | - when main nspawn supervisor process gets suspended due to SIGSTOP/SIGTTOU | |
2394 | or so, freeze the payload too. | |
2395 | - support time namespaces | |
2396 | - on cgroupsv1 issue cgroup empty handler process based on host events, so | |
2397 | that we make cgroup agent logic safe | |
2398 | - add API to invoke binary in container, then use that as fallback in | |
2399 | "machinectl shell" | |
2400 | - make nspawn suitable for shell pipelines: instead of triggering a hangup | |
2401 | when input is finished, send ^D, which synthesizes an EOF. Then wait for | |
2402 | hangup or ^D before passing on the EOF. | |
2403 | - greater control over selinux label? | |
2404 | - support that /proc, /sys/, /dev are pre-mounted | |
2405 | - maybe allow TPM passthrough, backed by swtpm, and measure --image= hash | |
2406 | into its PCR 11, so that nspawn instances can be TPM enabled, and partake | |
2407 | in measurements/remote attestation and such. swtpm would run outside of | |
2408 | control of container, and ideally would itself bind its encryption keys to | |
2409 | host TPM. | |
2410 | - make boot assessment do something sensible in a container. i.e send an | |
2411 | sd_notify() from payload to container manager once boot-up is completed | |
2412 | successfully, and use that in nspawn for dealing with boot counting, | |
2413 | implemented in the partition table labels and directory names. | |
2414 | - optionally set up nftables/iptables routes that forward UDP/TCP traffic on | |
2415 | port 53 to resolved stub 127.0.0.54 | |
2416 | - maybe optionally insert .nspawn file as GPT partition into images, so that | |
2417 | such container images are entirely stand-alone and can be updated as one. | |
fdeabf57 CB |
2418 | - The subreaper logic we currently have seems overly complex. We should |
2419 | investigate whether creating the inner child with CLONE_PARENT isn't better. | |
3922f982 CB |
2420 | - Reduce the number of sockets that are currently in use and just rely on one |
2421 | or two sockets. | |
3734fec7 | 2422 | - Support running nspawn as an unprivileged user. |
07eabc2b LB |
2423 | |
2424 | * machined: add API to acquire UID range. add API to mount/dissect loopback | |
2425 | file. Both protected by PK. Then make nspawn use these APIs to run | |
2426 | unprivileged containers. i.e. push the truly privileged bits into machined, | |
2427 | so that the client side can remain entirely unprivileged, with SUID or | |
2428 | anything like that. | |
2429 | ||
e25b5a8d | 2430 | * machined: |
e25b5a8d DH |
2431 | - add an API so that libvirt-lxc can inform us about network interfaces being |
2432 | removed or added to an existing machine | |
2433 | - "machinectl migrate" or similar to copy a container from or to a | |
2434 | difference host, via ssh | |
e25b5a8d DH |
2435 | - introduce systemd-nspawn-ephemeral@.service, and hook it into |
2436 | "machinectl start" with a new --ephemeral switch | |
2437 | - "machinectl status" should also show internal logs of the container in | |
2438 | question | |
e25b5a8d DH |
2439 | - "machinectl history" |
2440 | - "machinectl diff" | |
2441 | - "machinectl commit" that takes a writable snapshot of a tree, invokes a | |
2442 | shell in it, and marks it read-only after use | |
2443 | ||
abd55b16 | 2444 | * udev: |
abd55b16 | 2445 | - move to LGPL |
abd55b16 KS |
2446 | - kill scsi_id |
2447 | - add trigger --subsystem-match=usb/usb_device device | |
e8d842a0 | 2448 | - reimport udev db after MOVE events for devices without dev_t |
75723d31 | 2449 | - re-enable ProtectClock= once only cgroupsv2 is supported. |
8cfde28b | 2450 | See f562abe2963bad241d34e0b308e48cf114672c84. |
b8217b7b | 2451 | |
e25b5a8d DH |
2452 | * coredump: |
2453 | - save coredump in Windows/Mozilla minidump format | |
73a99163 | 2454 | - when truncating coredumps, also log the full size that the process had, and make a metadata field so we can report truncated coredumps |
1a0281a3 | 2455 | - add examples for other distros in ELF_PACKAGE_METADATA |
87a8baa3 LP |
2456 | |
2457 | * support crash reporting operation modes (https://live.gnome.org/GnomeOS/Design/Whiteboards/ProblemReporting) | |
2458 | ||
f38afcd0 | 2459 | * tmpfiles: |
f38afcd0 | 2460 | - apply "x" on "D" too (see patch from William Douglas) |
beca6b6e | 2461 | - allow time-based cleanup in r and R too |
614cc34f | 2462 | - instead of ignoring unknown fields, reject them. |
e25b5a8d DH |
2463 | - creating new directories/subvolumes/fifos/device nodes |
2464 | should not follow symlinks. None of the other adjustment or creation | |
2465 | calls follow symlinks. | |
36f57e02 | 2466 | - add --test mode |
ba405b22 ZJS |
2467 | - teach tmpfiles.d q/Q logic something sensible in the context of XFS/ext4 |
2468 | project quota | |
a9b0d0a2 | 2469 | - teach tmpfiles.d m/M to move / atomic move + symlink old -> new |
beca6b6e ZJS |
2470 | - add new line type for setting btrfs subvolume attributes (i.e. rw/ro) |
2471 | - tmpfiles: add new line type for setting fcaps | |
1258097c | 2472 | |
af6f0d42 TG |
2473 | * udev-link-config: |
2474 | - Make sure ID_PATH is always exported and complete for | |
2475 | network devices where possible, so we can safely rely | |
2476 | on Path= matching | |
af6f0d42 | 2477 | |
88e4d1d7 | 2478 | * sd-rtnl: |
88e4d1d7 | 2479 | - add support for more attribute types |
c589a0e6 | 2480 | - inbuilt piping support (essentially degenerate async)? see loopback-setup.c and other places |
88e4d1d7 | 2481 | |
0a4b9a07 | 2482 | * networkd: |
c74ecd71 TG |
2483 | - add more keys to [Route] and [Address] sections |
2484 | - add support for more DHCPv4 options (and, longer term, other kinds of dynamic config) | |
e8d842a0 | 2485 | - add reduced [Link] support to .network files |
798e174a | 2486 | - properly handle routerless dhcp leases |
a8eaaee7 | 2487 | - work with non-Ethernet devices |
e25b5a8d | 2488 | - dhcp: do we allow configuring dhcp routes on interfaces that are not the one we got the dhcp info from? |
e25b5a8d DH |
2489 | - the DHCP lease data (such as NTP/DNS) is still made available when |
2490 | a carrier is lost on a link. It should be removed instantly. | |
2491 | - expose in the API the following bits: | |
e2da6491 | 2492 | - option 15, domain name |
38b38500 | 2493 | - option 12, hostname and/or option 81, fqdn |
e25b5a8d DH |
2494 | - option 123, 144, geolocation |
2495 | - option 252, configure http proxy (PAC/wpad) | |
2496 | - provide a way to define a per-network interface default metric value | |
2497 | for all routes to it. possibly a second default for DHCP routes. | |
2498 | - allow Name= to be specified repeatedly in the [Match] section. Maybe also | |
2499 | support Name=foo*|bar*|baz ? | |
e25b5a8d | 2500 | - whenever uplink info changes, make DHCP server send out FORCERENEW |
155e8b9a | 2501 | |
07eabc2b LB |
2502 | * in networkd, when matching device types, fix up DEVTYPE rubbish the kernel passes to us |
2503 | ||
d5e172d2 ZJS |
2504 | * Figure out how to do unittests of networkd's state serialization |
2505 | ||
ac976532 | 2506 | * dhcp: |
424a8732 | 2507 | - figure out how much we can increase Maximum Message Size |
ac976532 | 2508 | |
37d8b536 PF |
2509 | * dhcp6: |
2510 | - add functions to set previously stored IPv6 addresses on startup and get | |
2511 | them at shutdown; store them in client->ia_na | |
2512 | - write more test cases | |
37d8b536 | 2513 | - implement reconfigure support, see 5.3., 15.11. and 22.20. |
37d8b536 | 2514 | - implement support for temporary adressess (IA_TA) |
37d8b536 PF |
2515 | - implement dhcpv6 authentication |
2516 | - investigate the usefulness of Confirm messages; i.e. are there any | |
2517 | situations where the link changes without any loss in carrier detection | |
2518 | or interface down | |
2519 | - some servers don't do rapid commit without a filled in IA_NA, verify | |
2520 | this behavior | |
4a77c53d | 2521 | - RouteTable= ? |
7af442cf MY |
2522 | |
2523 | * shared/wall: Once more programs are taught to prefer sd-login over utmp, | |
2524 | switch the default wall implementation to wall_logind | |
2525 | (https://github.com/systemd/systemd/pull/29051#issuecomment-1704917074) |