]> git.ipfire.org Git - thirdparty/systemd.git/log
thirdparty/systemd.git
2 years agoTEST-03-JOBS: test indirect PropagatesStopTo= 28262/head
Mike Yuan [Wed, 5 Jul 2023 18:40:01 +0000 (02:40 +0800)] 
TEST-03-JOBS: test indirect PropagatesStopTo=

2 years agocore/transaction: correctly skip unneeded operations for PropagatesStopTo=
Mike Yuan [Wed, 5 Jul 2023 18:17:36 +0000 (02:17 +0800)] 
core/transaction: correctly skip unneeded operations for PropagatesStopTo=

Follow-up for 48cb073db81fa73f64bc5aa9a1b81ebf627235fa

Break out from LIST_FOREACH correctly if nt == JOB_NOP.
Shouldn't have functional changes, just optimization.

2 years agoMerge pull request #28258 from poettering/boot-feature-catchup
Luca Boccassi [Wed, 5 Jul 2023 20:40:43 +0000 (21:40 +0100)] 
Merge pull request #28258 from poettering/boot-feature-catchup

sd-boot/sd-stub boot feature flag catchup

2 years agoMerge pull request #28255 from yuwata/sd-device-fix-clone
Luca Boccassi [Wed, 5 Jul 2023 20:38:52 +0000 (21:38 +0100)] 
Merge pull request #28255 from yuwata/sd-device-fix-clone

sd-device: fix device_clone_with_db()

2 years agotest: replace readfp() with read_file()
Frantisek Sumsal [Wed, 5 Jul 2023 17:43:43 +0000 (19:43 +0200)] 
test: replace readfp() with read_file()

ConfigParser.readfp() has been deprecated since Python 3.2 and was
dropped completely in Python 3.11.

2 years agohwdb update for v246-rc1
Lennart Poettering [Wed, 5 Jul 2023 20:00:19 +0000 (22:00 +0200)] 
hwdb update for v246-rc1

2 years agohostname-setup: don't pass "true" to a flags parameter
Lennart Poettering [Wed, 5 Jul 2023 16:02:52 +0000 (18:02 +0200)] 
hostname-setup: don't pass "true" to a flags parameter

2 years agoMerge pull request #28252 from yuwata/journal-open-machine
Mike Yuan [Wed, 5 Jul 2023 19:08:18 +0000 (03:08 +0800)] 
Merge pull request #28252 from yuwata/journal-open-machine

journal: introduce journal_open_machine()

2 years agoboot: make LoaderType enum less special 28258/head
Lennart Poettering [Wed, 5 Jul 2023 13:42:30 +0000 (15:42 +0200)] 
boot: make LoaderType enum less special

Usually (but not always) we use uppercase type naming, and do a typedef
for enums like this. Do so here too.

2 years agoboot: rename entry_count → n_entries
Lennart Poettering [Wed, 5 Jul 2023 13:41:26 +0000 (15:41 +0200)] 
boot: rename entry_count → n_entries

While we don't strictly follow the rule, most of our userspace names
these fields that count entries in some array n_xyz, hence let's do so
in the EFI boot code too, to make things less special.

2 years agoefi: add a bunch of reported EFI loader/stub feature flags
Lennart Poettering [Wed, 5 Jul 2023 13:38:38 +0000 (15:38 +0200)] 
efi: add a bunch of reported EFI loader/stub feature flags

We gained a bunch of new features that deserve reporting to userspace,
hence add matching flags for each.

This allows userspace to determine if installing addons in the ESP even
makes sense.

This is inspired by a similar changes in #28057

2 years agojournal-upload: add missing assertion 28252/head
Yu Watanabe [Wed, 5 Jul 2023 00:59:51 +0000 (09:59 +0900)] 
journal-upload: add missing assertion

2 years agojournal-upload: replace deprecated sd_journal_open_container()
Yu Watanabe [Wed, 5 Jul 2023 00:58:57 +0000 (09:58 +0900)] 
journal-upload: replace deprecated sd_journal_open_container()

2 years agojournal-util: extract journal_open_machine() from journalctl
Yu Watanabe [Wed, 5 Jul 2023 00:53:44 +0000 (09:53 +0900)] 
journal-util: extract journal_open_machine() from journalctl

2 years agosd-journal: introduce SD_JOURNAL_TAKE_DIRECTORY_FD flag for sd_journal_open_directory...
Yu Watanabe [Wed, 5 Jul 2023 00:49:45 +0000 (09:49 +0900)] 
sd-journal: introduce SD_JOURNAL_TAKE_DIRECTORY_FD flag for sd_journal_open_directory_fd()

If it is called with the flag, then the provided file descriptor will be
owned by the sd_journal object, and will be closed in sd_journal_close().

2 years agotest: change partition label to test if the outdated devlinks are removed 28255/head
Yu Watanabe [Wed, 5 Jul 2023 03:05:45 +0000 (12:05 +0900)] 
test: change partition label to test if the outdated devlinks are removed

The change is intended to reproduce the issue #27983, though the
original issue is highly racy, and the test does not reproduce it
reliably. But, anyway, it is better to change the partition label to
test the devlink removal.

2 years agosd-device: do not read uevent file in device_clone_with_db()
Yu Watanabe [Wed, 5 Jul 2023 02:44:00 +0000 (11:44 +0900)] 
sd-device: do not read uevent file in device_clone_with_db()

Follow-up for 381f6d4ba5551898e7ff19189485072b94879281.

When the function is called, the device may be already removed, and
another device has the same syspath. Such situation can occur when a
partition removed and another is created. In that case, the sysfs paths
of the removed and newly created partitions can be same, but their
devnums are different, and thus the database files corresponding to the
devices are also different.

Fixes #27981.

2 years agoMerge pull request #26844 from YHNdnzj/propagate-stop-fixup
Lennart Poettering [Wed, 5 Jul 2023 13:56:21 +0000 (15:56 +0200)] 
Merge pull request #26844 from YHNdnzj/propagate-stop-fixup

core: introduce UNIT_ATOM_PROPAGATE_STOP_GRACEFUL for PropagatesStopTo=

2 years agoMerge pull request #28207 from poettering/initrd-creds
Luca Boccassi [Wed, 5 Jul 2023 09:29:33 +0000 (10:29 +0100)] 
Merge pull request #28207 from poettering/initrd-creds

various credential improvements (including initrd creds, creds in generators, fstab + getty creds)

2 years agoMerge pull request #28253 from yuwata/hwdb-follow-up
Lennart Poettering [Wed, 5 Jul 2023 08:40:44 +0000 (10:40 +0200)] 
Merge pull request #28253 from yuwata/hwdb-follow-up

hwdb: several cleanups

2 years agoMerge pull request #28228 from yuwata/repart-free-area
Yu Watanabe [Wed, 5 Jul 2023 07:36:16 +0000 (16:36 +0900)] 
Merge pull request #28228 from yuwata/repart-free-area

repart: fix free area calculation

2 years agounit: also condition out systemd-backlight in initrd
Yu Watanabe [Wed, 5 Jul 2023 02:00:23 +0000 (11:00 +0900)] 
unit: also condition out systemd-backlight in initrd

Follow-up for 9173d31dfea5c2b05ff08480972c499cb7aac940.

The systemd-backlight@.service also save/restore state but the data
is in /var/.

2 years agounits: condition out a few services in the initrd
Lennart Poettering [Tue, 4 Jul 2023 20:31:11 +0000 (22:31 +0200)] 
units: condition out a few services in the initrd

Let's make our units more robust to being added to an initrd:

1. systemd-boot-update only makes sense if sd-boot is available in /usr/
   to copy into the ESP. This is generally not the case in initrds, and
   even if it was, we shouldn't update the ESP from the initrd, but from
   the host instead.

2. The rfkill services save/restore rfkill state, but that information
   is only available once /var/ is mounted, which generally happens
   after the initrd transition.

3. utmp management is partly in /var/, and legacy anyway, hence don't
   bother with it in the initrd.

2 years agotest: update test for free area calculation in repart 28228/head
Yu Watanabe [Mon, 3 Jul 2023 06:15:27 +0000 (15:15 +0900)] 
test: update test for free area calculation in repart

2 years agorepart: fix free area calculation
Yu Watanabe [Mon, 3 Jul 2023 06:14:14 +0000 (15:14 +0900)] 
repart: fix free area calculation

Like fdisk_get_last_lba(), fdisk_partition_get_end() return the last
sector in the partition.

Fixes #28225.

2 years agotest: add reproducer for issue #28225
Yu Watanabe [Mon, 3 Jul 2023 03:11:50 +0000 (12:11 +0900)] 
test: add reproducer for issue #28225

2 years agohwdb: make matching modalias for Archos 101 Cesium Educ more strict 28253/head
Yu Watanabe [Wed, 5 Jul 2023 01:17:59 +0000 (10:17 +0900)] 
hwdb: make matching modalias for Archos 101 Cesium Educ more strict

Follow-up for 41f34dcf3bc7004aaa9b251480d56bcfe144346d.

2 years agohwdb: merge multiple keyboard entries with same setting
Yu Watanabe [Wed, 5 Jul 2023 01:28:42 +0000 (10:28 +0900)] 
hwdb: merge multiple keyboard entries with same setting

Follow-up for 123c0e24dd3af4c7168ee77dad841cc730b5db60.

Note, the entry was originally added for IdeaPad Flex 5 in
21b589a15504cdbd309a82abb566ef4e36957f92.
Then, a bug introduced by 19db450f3a243fcaf0949beebafc3025f8e3a98e.
But, when it was fixed by 738a195bd59dc2e85c15382d17d391c1781aaa4e,
the glob becomes too stricter, and another variant was added by
123c0e24dd3af4c7168ee77dad841cc730b5db60.

2 years agohwdb: drop trailing white space
Yu Watanabe [Wed, 5 Jul 2023 01:16:56 +0000 (10:16 +0900)] 
hwdb: drop trailing white space

Follow-up for 4687f001e689fd482f530a8d1d0adc93e01d74ea.

2 years agojournal-upload: make --namespace=* work
Yu Watanabe [Wed, 5 Jul 2023 00:19:16 +0000 (09:19 +0900)] 
journal-upload: make --namespace=* work

Follow-up for 9f6e0bd417fa287dd1e7b541bfe0c60f04cc29e4.

Note that sd_journal_open() is a simple wrapper of sd_journal_open_namespace(),
hence we can merge the two branch.

2 years agotest: add test cases for issue #26839 in TEST-03-JOBS 26844/head
Mike Yuan [Thu, 16 Mar 2023 08:23:17 +0000 (16:23 +0800)] 
test: add test cases for issue #26839 in TEST-03-JOBS

2 years agocore: introduce UNIT_ATOM_PROPAGATE_STOP_GRACEFUL for PropagatesStopTo=
Mike Yuan [Mon, 3 Jul 2023 23:28:33 +0000 (07:28 +0800)] 
core: introduce UNIT_ATOM_PROPAGATE_STOP_GRACEFUL for PropagatesStopTo=

Follow-up for 017a7ba4f406adcf69d6b3ec15b9f2d9ed5ad853

Before this commit, when a unit that is restarting propagates stop
to other units, it can also depend on them, which results in
job type conflict and thus failure to pull in the dependencies.

So, let's introduce a new dependency atom UNIT_ATOM_PROPAGATE_STOP_GRACEFUL,
and use it for PropagatesStopTo=. It will enqueue a restart job if
there's already a start job, which meets the ultimate goal and avoids
job type conflict.

Fixes #26839

2 years agocore/transaction: return early when appropriate to reduce nesting
Mike Yuan [Mon, 3 Jul 2023 22:55:55 +0000 (06:55 +0800)] 
core/transaction: return early when appropriate to reduce nesting

2 years agoman/systemctl: document --job-mode=restart-dependencies
Mike Yuan [Wed, 5 Jul 2023 00:14:27 +0000 (08:14 +0800)] 
man/systemctl: document --job-mode=restart-dependencies

Follow-up for 09d04ad325473e05e23e6ba8382d7de1dd819bda

2 years agoupdate TODO 28207/head
Lennart Poettering [Fri, 23 Jun 2023 20:50:02 +0000 (22:50 +0200)] 
update TODO

2 years agoimport-creds: add support for binary credentials specified on the kernel cmdline
Lennart Poettering [Tue, 4 Jul 2023 09:46:37 +0000 (11:46 +0200)] 
import-creds: add support for binary credentials specified on the kernel cmdline

2 years agoman: document where PID 1 imports credentials from
Lennart Poettering [Fri, 30 Jun 2023 09:46:16 +0000 (11:46 +0200)] 
man: document where PID 1 imports credentials from

2 years agoman: document how credentials are passed into generators
Lennart Poettering [Fri, 30 Jun 2023 09:45:59 +0000 (11:45 +0200)] 
man: document how credentials are passed into generators

2 years agodoc: document inird credentials + and how to consume credentials in generators
Lennart Poettering [Fri, 30 Jun 2023 09:44:46 +0000 (11:44 +0200)] 
doc: document inird credentials + and how to consume credentials in generators

(as well as various other fixes)

2 years agogenerator: run various generators only run on the host, not in initrd
Lennart Poettering [Fri, 30 Jun 2023 08:23:10 +0000 (10:23 +0200)] 
generator: run various generators only run on the host, not in initrd

These 5 generators only make sense on the host,not in the initrd, hence
if they end up in the initrd anyway, make them exit quickly.

2 years agotest: verify that the getty generator with creds works
Lennart Poettering [Thu, 29 Jun 2023 20:53:26 +0000 (22:53 +0200)] 
test: verify that the getty generator with creds works

2 years agogetty-generator: minor modernizations
Lennart Poettering [Fri, 23 Jun 2023 20:50:55 +0000 (22:50 +0200)] 
getty-generator: minor modernizations

2 years agogetty-generator: allow configuring additional gettys via credentials
Lennart Poettering [Fri, 23 Jun 2023 20:50:34 +0000 (22:50 +0200)] 
getty-generator: allow configuring additional gettys via credentials

2 years agotest: add simple fstab credential test
Lennart Poettering [Thu, 29 Jun 2023 19:52:57 +0000 (21:52 +0200)] 
test: add simple fstab credential test

2 years agofstab-generator: add more parameter name comments
Lennart Poettering [Fri, 23 Jun 2023 20:49:55 +0000 (22:49 +0200)] 
fstab-generator: add more parameter name comments

2 years agofstab-generator: optional read addtional fstab lines from credentials
Lennart Poettering [Fri, 23 Jun 2023 19:51:12 +0000 (21:51 +0200)] 
fstab-generator: optional read addtional fstab lines from credentials

Fixes: #27260
2 years agotest: add test for initrd credentials
Lennart Poettering [Thu, 29 Jun 2023 17:03:08 +0000 (19:03 +0200)] 
test: add test for initrd credentials

This extends the test framework a bit, and allows adding additional
initrds to the qemu invocation, which we use here to place credentials
in the new /run/systemd/@initrd/ credentials dir which are then passed
to the host.

2 years agoimport-creds: unify acquire_credential_directory() + acquire_encrypted_credential_dir...
Lennart Poettering [Thu, 29 Jun 2023 10:53:23 +0000 (12:53 +0200)] 
import-creds: unify acquire_credential_directory() + acquire_encrypted_credential_directory()

Let's unify these very similar functions, and port them to the new
mount_credentials_fs() call.

While we are at it, if we detect that the credentials dir already is a
mount point, remount it writable so that we can actually write to it.

2 years agoexecute: split out mounting of credentials fs
Lennart Poettering [Thu, 29 Jun 2023 10:32:44 +0000 (12:32 +0200)] 
execute: split out mounting of credentials fs

Let's add two new helpers: mount_credentials_fs() and
credentials_fs_mount_flags(). The former mounts a file system suitable
for storing of unencrypted credentials at runtime (i.e. a ramfs or
tmpfs). The latter determines the right mount flags to use for such a
mount.

Both functions mostly just take code from execute.c, but make two
changes:

1. If the kernel supports it we'll use a tmpfs with the new "noswap"
   mount option instead of ramfs. Was added in kernel 6.4, hence is very
   recent, but tmpfs is so much less crappy than ramfs, hence worth it.

2. We'll set MS_NOSYMFOLLOW on the mounts if supported. These file
   systems should only contain regulra files, hence no need to allow
   symlinks.

2 years agoimport-creds: show list of imported credentials during initialization of PID 1
Lennart Poettering [Wed, 28 Jun 2023 20:58:07 +0000 (22:58 +0200)] 
import-creds: show list of imported credentials during initialization of PID 1

Let's make things easier to debug: provide an overview what has been
passed, during boot.

2 years agocore: consult credentials for machine ID to use for host
Lennart Poettering [Wed, 28 Jun 2023 16:11:15 +0000 (18:11 +0200)] 
core: consult credentials for machine ID to use for host

Let's hook up one more thing with credentials: the machine ID to use
when none is initialized yet.

This requires some reordering of initialization steps in PID 1: we need
to import credentials first, and only then initialize the machine ID.

2 years agoimport-creds: pick up vmm.notify_socket also from encrypted credentials
Lennart Poettering [Wed, 28 Jun 2023 16:10:40 +0000 (18:10 +0200)] 
import-creds: pick up vmm.notify_socket also from encrypted credentials

Now that we have the infra in place, make PID 1 pick up encrypted
credentials too.

(While we are at it, split this out into its own helper)

2 years agocreds-util: add new helper read_credential_with_decryption()
Lennart Poettering [Fri, 23 Jun 2023 19:48:15 +0000 (21:48 +0200)] 
creds-util: add new helper read_credential_with_decryption()

This is just like read_credential() but also looks into the encrypted
credential directory, not just the regular one.

Normally, we decrypt credentials at the moment we pass them to services.
From service PoV all credentials are hence decrypted credentials.

However, when we want to access credentials in a generator this logic
does not apply: here we have the regular and the encrypted credentials
directory. So far we didn't attempt to make use of credentials in
generators hence.

Let's address and add helper that looks into both directories, and talks
to the TPM if necessary to decrypt the credentials.

2 years agoimport-creds: define a new dir where initrd configurators can pass credentials to...
Lennart Poettering [Fri, 23 Jun 2023 15:49:44 +0000 (17:49 +0200)] 
import-creds: define a new dir where initrd configurators can pass credentials to host

2 years agoexecute: fix credential dir handling for fs which support ACLs
Lennart Poettering [Tue, 4 Jul 2023 20:26:52 +0000 (22:26 +0200)] 
execute: fix credential dir handling for fs which support ACLs

When the credential dir is backed by an fs that supports ACLs we must be
more careful with adjusting the 'x' bit of the directory, as any chmod()
call on the dir will reset the mask entry of the ACL entirely which we
don't want. Hence, do a manual set of ACL changes, that only add/drop
the 'x' bit but otherwise leave the ACL as it is.

This matters if we use tmpfs rather than ramfs to store credentials.

2 years agoman: make sure credentials properly show up in directives index
Lennart Poettering [Fri, 30 Jun 2023 08:22:35 +0000 (10:22 +0200)] 
man: make sure credentials properly show up in directives index

2 years agojournal-remote: upload journals from namespace
Igor Tsiglyar [Mon, 3 Jul 2023 13:51:32 +0000 (16:51 +0300)] 
journal-remote: upload journals from namespace

2 years agohwdb : add support for Archos 101 Cesium Educ to 60-sensor.hwdb
Thomas Genty [Tue, 4 Jul 2023 13:41:57 +0000 (15:41 +0200)] 
hwdb : add support for Archos 101 Cesium Educ to 60-sensor.hwdb

2 years agoNEWS/--help: correct/be clearer on bootclt -R vs. -RR
Lennart Poettering [Tue, 4 Jul 2023 11:04:56 +0000 (13:04 +0200)] 
NEWS/--help: correct/be clearer on bootclt -R vs. -RR

The NEWS file was simply wrong. Let's also improve the --help text on
this.

Fixes: #28221
2 years agohibernate-resume-generator: downgrade 'noresume' log message
Lennart Poettering [Tue, 4 Jul 2023 09:47:19 +0000 (11:47 +0200)] 
hibernate-resume-generator: downgrade 'noresume' log message

This log message is shown pretty regular at boot in various scenarios
(such as CI builds), and it's not a reason for any concern, it's just the
immediate effect of explicit configuration. Hence let's downgrade from
LOG_NOTICE to LOG_INFO so that it is still usually in the boot output,
but not particularly highlighted, since there's really no reason to.

2 years agosystemctl: implement a new "whoami" verb, that just returns unit of caller/PID
Lennart Poettering [Mon, 3 Jul 2023 12:36:14 +0000 (14:36 +0200)] 
systemctl: implement a new "whoami" verb, that just returns unit of caller/PID

2 years agotest: wait for the interface to become routable after reconfiguring
Frantisek Sumsal [Mon, 3 Jul 2023 19:38:42 +0000 (21:38 +0200)] 
test: wait for the interface to become routable after reconfiguring

Since 6e8477edd3 TEST-75 started failing with:

[  571.468298] testsuite-75.sh[46]: + for addr in "${DNS_ADDRESSES[@]}"
[  571.468298] testsuite-75.sh[46]: + run delv @fd00:dead:beef:cafe::1 -t A mail.signed.test
[  571.468899] testsuite-75.sh[562]: + tee /tmp/tmp.qKlHPbCCJZ
[  571.469317] testsuite-75.sh[561]: + delv @fd00:dead:beef:cafe::1 -t A mail.signed.test
[  571.501381] testsuite-75.sh[562]: ;; network unreachable resolving 'mail.signed.test/A/IN': fd00:dead:beef:cafe::1#53
[  571.501564] testsuite-75.sh[562]: ;; resolution failed: SERVFAIL
[  571.515457] testsuite-75.sh[46]: + grep -qF '; fully validated' /tmp/tmp.qKlHPbCCJZ

Let's wait for the dns0 interface to become routable again after
re-enabling IPv6 to, hopefully, mitigate this.

2 years agoMerge pull request #28245 from rpigott/dhcp6-relax-assert
Yu Watanabe [Mon, 3 Jul 2023 23:06:13 +0000 (08:06 +0900)] 
Merge pull request #28245 from rpigott/dhcp6-relax-assert

relax data assert in dchp6_option_parse_string

2 years agoMerge pull request #28215 from poettering/start-as-restart
Mike Yuan [Mon, 3 Jul 2023 21:54:56 +0000 (05:54 +0800)] 
Merge pull request #28215 from poettering/start-as-restart

pid1: enqueue auto-restart job for the unit itself as JOB_START and for dependending units as JOB_RESTART

2 years agotest-network: add test for an invalid captive portal uri 28245/head
Ronan Pigott [Mon, 3 Jul 2023 20:15:47 +0000 (13:15 -0700)] 
test-network: add test for an invalid captive portal uri

This could probably be extended to include many more invalid uri

2 years agodhcp6: relax data assert in dhcp6_option_parse_string
Ronan Pigott [Mon, 3 Jul 2023 20:02:47 +0000 (13:02 -0700)] 
dhcp6: relax data assert in dhcp6_option_parse_string

dhcp6_option_parse_string is intended to clear strings with length 0,
for consistency. The data assert is too strict for this purpose, so we
will allow data || data_len == 0, similar to other dhcp6_option_parse*
helpers.

Fixes: fde788601be8 ("dhcp6-client: parse RFC8910 captive portal dhcp6 option")
2 years agoMerge pull request #28232 from rpigott/ndisc-captive-portal-mfree
Yu Watanabe [Mon, 3 Jul 2023 16:27:20 +0000 (01:27 +0900)] 
Merge pull request #28232 from rpigott/ndisc-captive-portal-mfree

ndisc: clear ndisc captive portal value on bogus zero-len option

2 years agorepart: fix comment
Yu Watanabe [Mon, 3 Jul 2023 05:37:07 +0000 (14:37 +0900)] 
repart: fix comment

2 years agologinctl,machinectl: use same ansi formatting in --help texts as in systemctl
Lennart Poettering [Mon, 3 Jul 2023 11:33:47 +0000 (13:33 +0200)] 
loginctl,machinectl: use same ansi formatting in --help texts as in systemctl

2 years agounits: skip systemd-battery-check in environments where it doesn't make sense
Lennart Poettering [Mon, 3 Jul 2023 11:36:08 +0000 (13:36 +0200)] 
units: skip systemd-battery-check in environments where it doesn't make sense

Let's condition the service so that it doesn't run where we aren't
directly run on baremetal, or where no power sources are discovered at
all.

2 years agocore/service: show correct restart usec for services in SERVICE_AUTO_RESTART_QUEUED 28215/head
Mike Yuan [Mon, 3 Jul 2023 14:32:36 +0000 (22:32 +0800)] 
core/service: show correct restart usec for services in SERVICE_AUTO_RESTART_QUEUED

Follow-up for #28215

We can now correctly distinguish enqueued auto-restarts
from those that are still pending. Let's take advantage
of that.

2 years agotest: add test case for recent OnSuccess=/OnFailure= state machine changes
Lennart Poettering [Mon, 3 Jul 2023 12:49:46 +0000 (14:49 +0200)] 
test: add test case for recent OnSuccess=/OnFailure= state machine changes

2 years agocore: introduce a new job mode JOB_RESTART_DEPENDENCIES
Lennart Poettering [Fri, 30 Jun 2023 16:17:06 +0000 (18:17 +0200)] 
core: introduce a new job mode JOB_RESTART_DEPENDENCIES

This new job mode will enqueue a start job for a unit, and all units
depending on the unit will get a restart job enqueued. This is then used
for automatic sevice restarts: the unit itself is only started, the
depending units restarted. This way the unit will not go down
unnecessarily, triggering OnSuccess= needlessly.

This also introduces a new state SERVICE_AUTO_RESTART_QUEUED that is
entered once the restart jobs are enqueued. Previously we'd stay in
SERVICE_AUTO_RESTART, but that's problematic, since we'd lose
information whether we still need to enqueue the restart job during a
serialization/deserialization cycle or not. By having an explicit state
for this we know exactly whether we still need to enqueue the job or
not. It's also good since when we are in SERVICE_AUTO_RESTART_QUEUED we
want to act on unit_start(), but on SERVICE_AUTO_RESTART we want to wait
for the holdoff time to pass before we act on unit_start().

Fixes: #27722
2 years agoservice: re-linebreak some comments matching current coding style
Lennart Poettering [Fri, 30 Jun 2023 16:16:55 +0000 (18:16 +0200)] 
service: re-linebreak some comments matching current coding style

2 years agoMerge pull request #28230 from yuwata/network-wait-address-configure
Luca Boccassi [Mon, 3 Jul 2023 14:04:32 +0000 (15:04 +0100)] 
Merge pull request #28230 from yuwata/network-wait-address-configure

network: delay to configure address untill it is removed on reconfigure

2 years agondisc: reject malformed captive portal URI with EBADMSG 28232/head
Ronan Pigott [Mon, 3 Jul 2023 10:07:57 +0000 (03:07 -0700)] 
ndisc: reject malformed captive portal URI with EBADMSG

This allows the correct, gracious, error handling to follow up in the
ndisc handler. Otherwise, an internal error is assumed and the interface
disabled.

Fixes: 9747955d2d60 ("ndisc: parse RFC8910 captive portal ipv6ra option")
2 years agoupdate TODO
Lennart Poettering [Mon, 3 Jul 2023 09:59:45 +0000 (11:59 +0200)] 
update TODO

2 years agondisc: clear ndisc captive portal value on bogus zero-len option
Ronan Pigott [Mon, 3 Jul 2023 07:36:04 +0000 (00:36 -0700)] 
ndisc: clear ndisc captive portal value on bogus zero-len option

This value was freed but erroneously never cleared, leading to
use-after-free.

Fixes: 9747955d2d60 ("ndisc: parse RFC8910 captive portal ipv6ra option")
2 years agotest-network: add test for static route with preferred source 28230/head
Yu Watanabe [Mon, 3 Jul 2023 07:04:30 +0000 (16:04 +0900)] 
test-network: add test for static route with preferred source

This adds possible reproducer for issue #28009 (though, the issue is
highly racy, hence this may not trigger the issue reliably).

2 years agotest-network: check route more strictly
Yu Watanabe [Mon, 3 Jul 2023 07:03:50 +0000 (16:03 +0900)] 
test-network: check route more strictly

2 years agonetwork: delay to configure address until it is removed on reconfigure
Yu Watanabe [Mon, 3 Jul 2023 06:43:53 +0000 (15:43 +0900)] 
network: delay to configure address until it is removed on reconfigure

When we request an address that already exists and is under removing,
we need to wait for the address being removed. Otherwise, configuration
of a route whose preferred source is the address will fail.

Fixes #28009.
Replaces #28088.

2 years agonetwork: constify several functions
Yu Watanabe [Mon, 3 Jul 2023 07:00:20 +0000 (16:00 +0900)] 
network: constify several functions

2 years agoMerge pull request #28132 from rpigott/dhcp-captive-portal
Yu Watanabe [Mon, 3 Jul 2023 05:51:56 +0000 (14:51 +0900)] 
Merge pull request #28132 from rpigott/dhcp-captive-portal

Implement RFC8910: captive portal dhcp options

2 years agohwdb: Add override for headset form-factors
Bastien Nocera [Sun, 2 Jul 2023 09:44:57 +0000 (11:44 +0200)] 
hwdb: Add override for headset form-factors

Correct the SOUND_FORM_FACTOR property for Steelseries Arctis headsets.
The USB IDs were all gathered from HeadsetControl[1].

[1]: https://github.com/Sapd/HeadsetControl

2 years agoMerge pull request #28226 from mrc0mmand/coverity-fixes
Luca Boccassi [Sun, 2 Jul 2023 14:54:56 +0000 (15:54 +0100)] 
Merge pull request #28226 from mrc0mmand/coverity-fixes

A couple of Coverity-related fixes

2 years agotree-wide: explicitly ignore return value in a couple more places 28226/head
Frantisek Sumsal [Sun, 2 Jul 2023 09:39:00 +0000 (11:39 +0200)] 
tree-wide: explicitly ignore return value in a couple more places

Resolves:
  - CID#1490777
  - CID#1498366
  - CID#1508639
  - CID#1509084
  - CID#1509086
  - CID#1509087

2 years agounit: drop an unused assignment
Frantisek Sumsal [Sun, 2 Jul 2023 09:22:50 +0000 (11:22 +0200)] 
unit: drop an unused assignment

Resolves: CID#1509244

2 years agovirt: drop an unused assignment
Frantisek Sumsal [Sun, 2 Jul 2023 09:09:09 +0000 (11:09 +0200)] 
virt: drop an unused assignment

As `v` gets overwritten by the following detect_vm_cpuid() call.

Resolves: CID#1509247

2 years agotree-wide: "<n>bit" → "<n>-bit"
Zbigniew Jędrzejewski-Szmek [Sat, 1 Jul 2023 21:33:20 +0000 (15:33 -0600)] 
tree-wide: "<n>bit" → "<n>-bit"

In some places, "<n> bits" is used when more appropriate.

2 years agotest: drop an unused assignment
Frantisek Sumsal [Sun, 2 Jul 2023 09:04:38 +0000 (11:04 +0200)] 
test: drop an unused assignment

Unused since 788c2d9523.
Resolves: CID#1509248

2 years agotest-network: add tests for captive portal dhcp options 28132/head
Ronan Pigott [Thu, 29 Jun 2023 23:38:55 +0000 (16:38 -0700)] 
test-network: add tests for captive portal dhcp options

2 years agonetworkctl: show captive portal configuration in link status
Ronan Pigott [Thu, 29 Jun 2023 23:38:26 +0000 (16:38 -0700)] 
networkctl: show captive portal configuration in link status

2 years agonetworkd: include captive portal information in link json description
Ronan Pigott [Thu, 29 Jun 2023 23:36:50 +0000 (16:36 -0700)] 
networkd: include captive portal information in link json description

2 years agonetwork: Introduce UseCaptivePortal IPv6RA option
Ronan Pigott [Thu, 29 Jun 2023 23:58:03 +0000 (16:58 -0700)] 
network: Introduce UseCaptivePortal IPv6RA option

Accepts a boolean. When enabled retains captive portal configuration
advertised by the router.

2 years agonetwork: Introduce UseCaptivePortal DHCPv6 option
Ronan Pigott [Thu, 29 Jun 2023 23:33:57 +0000 (16:33 -0700)] 
network: Introduce UseCaptivePortal DHCPv6 option

Acepts a boolean. When enabled requests and retains captive portal
configuration from the DHCPv6 server.

2 years agonetwork: Introduce UseCaptivePortal DHCPv4 option
Ronan Pigott [Thu, 29 Jun 2023 23:30:31 +0000 (16:30 -0700)] 
network: Introduce UseCaptivePortal DHCPv4 option

Accepts a boolean. When enabled, UseCaptivePortal will request and
retain the captive portal configuration from the DHCP server.

2 years agondisc: parse RFC8910 captive portal ipv6ra option
Ronan Pigott [Thu, 29 Jun 2023 23:22:45 +0000 (16:22 -0700)] 
ndisc: parse RFC8910 captive portal ipv6ra option

2 years agoNEWS: reword/fix/extend the entries for v254
Zbigniew Jędrzejewski-Szmek [Sat, 1 Jul 2023 20:07:25 +0000 (14:07 -0600)] 
NEWS: reword/fix/extend the entries for v254

2 years agoNEWS: drop doubled space after period
Zbigniew Jędrzejewski-Szmek [Sat, 1 Jul 2023 19:54:14 +0000 (13:54 -0600)] 
NEWS: drop doubled space after period

Emacs does this, but it doesn't make much sense. It makes even
less sense if done only in some places.

2 years agobuild(deps): bump github/codeql-action from 2.3.5 to 2.20.1
dependabot[bot] [Sat, 1 Jul 2023 10:38:05 +0000 (10:38 +0000)] 
build(deps): bump github/codeql-action from 2.3.5 to 2.20.1

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 2.3.5 to 2.20.1.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/0225834cc549ee0ca93cb085b92954821a145866...f6e388ebf0efc915c6c5b165b019ee61a6746a38)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>