]> git.ipfire.org Git - thirdparty/systemd.git/log
thirdparty/systemd.git
6 weeks agoMerge pull request #32709 from bluca/machined_ssh
Luca Boccassi [Wed, 8 May 2024 12:37:55 +0000 (14:37 +0200)] 
Merge pull request #32709 from bluca/machined_ssh

machined: add GetMachineSSHInfo method and varlink interface to register machines

6 weeks agoMerge pull request #32708 from DaanDeMeyer/fix-race
Luca Boccassi [Wed, 8 May 2024 12:21:37 +0000 (14:21 +0200)] 
Merge pull request #32708 from DaanDeMeyer/fix-race

Fix race conditions in tests

6 weeks agohwdb: Add a Logitech MX Master 3S (connected via Bolt Receiver)
Maciej S. Szmigiero [Tue, 7 May 2024 20:51:46 +0000 (22:51 +0200)] 
hwdb: Add a Logitech MX Master 3S (connected via Bolt Receiver)

The mouse technically has settable DPI between 200 and 8000, with a step of
50, but 1000 DPI is the default setting.
And if someone customizes the mouse DPI setting they can make a custom hwdb
rule too.

6 weeks agoMerge pull request #32705 from YHNdnzj/hibernate-error
Luca Boccassi [Wed, 8 May 2024 12:19:07 +0000 (14:19 +0200)] 
Merge pull request #32705 from YHNdnzj/hibernate-error

hibernate-util: differentiate some errors from the generic ENOSPC; systemctl: adjust the fallback behavior for sleep operations

6 weeks agoMerge pull request #32710 from YHNdnzj/debug-generator-cleanup
Daan De Meyer [Wed, 8 May 2024 12:09:56 +0000 (14:09 +0200)] 
Merge pull request #32710 from YHNdnzj/debug-generator-cleanup

debug-generator: several cleanups

6 weeks agoMerge pull request #32234 from poettering/dlopen-name-elf-note
Luca Boccassi [Wed, 8 May 2024 11:56:34 +0000 (13:56 +0200)] 
Merge pull request #32234 from poettering/dlopen-name-elf-note

Expose dlopen() dependencies in an ELF section, and add spec for it

6 weeks agoUse the correct name of CEL
MaxHearnden [Wed, 8 May 2024 06:02:52 +0000 (07:02 +0100)] 
Use the correct name of CEL

It's Canonical Event Format, not Common

6 weeks agomachined: add varlink interface for registering machines 32709/head
Sam Leonard [Tue, 23 Apr 2024 09:26:59 +0000 (10:26 +0100)] 
machined: add varlink interface for registering machines

This commit adds the new varlink interface io.systemd.Machine at
/run/systemd/machine/io.systemd.Machine with a single method Register

It supports all combinations of RegisterMachine[WithSSH,WithNetwork] all
under the same method.

6 weeks agodebug-generator: clean up logging and such for units passed from creds 32710/head
Mike Yuan [Wed, 8 May 2024 10:18:33 +0000 (18:18 +0800)] 
debug-generator: clean up logging and such for units passed from creds

6 weeks agodebug-generator: don't eat up error if we can't write drop-in
Mike Yuan [Wed, 8 May 2024 10:13:29 +0000 (18:13 +0800)] 
debug-generator: don't eat up error if we can't write drop-in

While at it, prepend the drop-in text with
our usual "Automatically generated by ...".

6 weeks agodebug-generator: drop unneeded empty_to_root
Mike Yuan [Wed, 8 May 2024 10:41:54 +0000 (18:41 +0800)] 
debug-generator: drop unneeded empty_to_root

6 weeks agodebug-generator: use RET_GATHER where appropriate
Mike Yuan [Wed, 8 May 2024 10:05:15 +0000 (18:05 +0800)] 
debug-generator: use RET_GATHER where appropriate

6 weeks agodebug-generator: enclose function call in if statement when retval
Mike Yuan [Wed, 8 May 2024 09:57:28 +0000 (17:57 +0800)] 
debug-generator: enclose function call in if statement when retval
is not used

6 weeks agoTEST-04-JOURNAL: Sleep more in delegated cgroup filtering script 32708/head
Daan De Meyer [Wed, 8 May 2024 09:44:28 +0000 (11:44 +0200)] 
TEST-04-JOURNAL: Sleep more in delegated cgroup filtering script

We already changed logs-filtering.service to sleep 2 seconds before
exiting to combat flakyness, let's do the same for the delegated
cgroup filtering payload.

Fixes #32696 (hopefully)

6 weeks agoTEST-81-GENERATORS: Do a lazy unmounts
Daan De Meyer [Wed, 8 May 2024 09:41:04 +0000 (11:41 +0200)] 
TEST-81-GENERATORS: Do a lazy unmounts

Otherwise we might fail if PID 1 is currently accessing these files.

Fixes #32692 (hopefully)

6 weeks agossh-generator: fix pointer cast type
Mike Yuan [Wed, 8 May 2024 10:12:35 +0000 (18:12 +0800)] 
ssh-generator: fix pointer cast type

6 weeks agotree-wide: add dlopen ELF notes to all dlopen() deps of ours 32234/head
Lennart Poettering [Thu, 11 Apr 2024 15:50:41 +0000 (17:50 +0200)] 
tree-wide: add dlopen ELF notes to all dlopen() deps of ours

Use 'recommended' priority for the default compression library, to
indicate that it should be prioritized over the other ones, as it
will be used to compress journals/core files.
Also use 'recommended' for kmod, as systems will likely fail to boot
if it's missing from the initrd.
Use 'suggested' for everything else.

There is one dlopen'ed TPM library that has the name generated
at runtime (depending on the driver), so that cannot be added, as it
needs to be known at build time.
Also when we support multiple ABI versions list them all, as for the
same reason we cannot know which one will be used at build time.

$ dlopen-notes.py build/libsystemd.so.0.39.0 build/src/shared/libsystemd-shared-256.so
libarchive.so.13 suggested
libbpf.so.0 suggested
libbpf.so.1 suggested
libcryptsetup.so.12 suggested
libdw.so.1 suggested
libelf.so.1 suggested
libfido2.so.1 suggested
libgcrypt.so.20 suggested
libidn2.so.0 suggested
libip4tc.so.2 suggested
libkmod.so.2 recommended
liblz4.so.1 suggested
liblzma.so.5 suggested
libp11-kit.so.0 suggested
libpcre2-8.so.0 suggested
libpwquality.so.1 suggested
libqrencode.so.3 suggested
libqrencode.so.4 suggested
libtss2-esys.so.0 suggested
libtss2-mu.so.0 suggested
libtss2-rc.so.0 suggested
libzstd.so.1 recommended

Co-authored-by: Luca Boccassi <bluca@debian.org>
6 weeks agodlfcn: add macro for exporting dlopen() module names in ELF sections
Lennart Poettering [Thu, 11 Apr 2024 15:45:17 +0000 (17:45 +0200)] 
dlfcn: add macro for exporting dlopen() module names in ELF sections

This allows code to declare "weak" dlopen() style deps via an ELF
section following the just added specification.

The idea is that any user of dlopen() will place ELF_NOTE_DLOPEN(…)
somewhere close which will synthesize the note.

Tools such as rpm/dpkg package builders as well as initrd generators
(such as dracut) can then automatically pick up these weak deps of
suggested dependencies for their purposes.

Co-authored-by: Luca Boccassi <bluca@debian.org>
6 weeks agodocs: add specification for ELF dlopen metadata
Luca Boccassi [Sun, 5 May 2024 15:45:21 +0000 (16:45 +0100)] 
docs: add specification for ELF dlopen metadata

Tool to parse it and to use it for Debian packaging available at
https://github.com/systemd/package-notes

Add 3 levels of priority like Debian does, but using terminology
from RPM as it seems more apt.
We will very likely never use 'required', as we use this functionality
for optional features, but it is worth having it in the spec nonetheless
as we want it to be generalized.

6 weeks agotmpfiles: Don't fail if file does not exist in item_do()
Daan De Meyer [Wed, 8 May 2024 09:35:21 +0000 (11:35 +0200)] 
tmpfiles: Don't fail if file does not exist in item_do()

If the file was removed by some other program, we should just go
to the next one without failing. item_do() is only used for recursive
globs instead of fixed paths so skipping on missing files makes sense
(unlike if the path was fixed where we should probably fail).

Fixes #32691 (hopefully)

6 weeks agoTEST-74-AUX-UTILS: Make more robust
Daan De Meyer [Wed, 8 May 2024 09:21:42 +0000 (11:21 +0200)] 
TEST-74-AUX-UTILS: Make more robust

Let's run mkfs on the file we create instead of the loop device and
let's use udevadm wait --settle to wait for udev to settle before
doing anything with the loop device

Fixes #32680 (hopefully)

6 weeks agomachined: add GetMachineSSHInfo method
Sam Leonard [Wed, 24 Apr 2024 17:08:13 +0000 (18:08 +0100)] 
machined: add GetMachineSSHInfo method

Also adds three properties:
- VsockCid: the VSOCK CID of the VM
- SshAddress: the address of the VM in a format SSH can connect to
- SshPrivateKeyPath: the path to the SSH private key to use to connect
  to the VM.

GetMachineSSHInfo is essentially a convenience method to query both the
SshAddress and SshPrivateKeyPath properties at once.

6 weeks agomachined: split manager linking out of machine_new into machine_link
Sam Leonard [Wed, 24 Apr 2024 08:57:18 +0000 (09:57 +0100)] 
machined: split manager linking out of machine_new into machine_link

6 weeks agomachined: expose machine_freep in machine.h
Sam Leonard [Mon, 22 Apr 2024 15:23:13 +0000 (16:23 +0100)] 
machined: expose machine_freep in machine.h

6 weeks agoshared/json: add json_dispatch_absolute_path
Sam Leonard [Tue, 23 Apr 2024 14:03:34 +0000 (15:03 +0100)] 
shared/json: add json_dispatch_absolute_path

6 weeks agoTEST-07-PID1: Fix race in aux-scope subtest
Daan De Meyer [Wed, 8 May 2024 08:20:20 +0000 (10:20 +0200)] 
TEST-07-PID1: Fix race in aux-scope subtest

Currently test-aux-scope.service can get killed by the test before
it's had a chance to setup its signal handler. Make it Type=notify
to fix the race.

Fixes #32670 (hopefully)

6 weeks agoman/run0: remove the --user example for --machine=
Mike Yuan [Wed, 8 May 2024 05:55:00 +0000 (13:55 +0800)] 
man/run0: remove the --user example for --machine=

run0's --user= option is different from other tools,
and the whole point of run0 is to connect to the system
manager. So the example is spurious.

6 weeks agopreset-all: continue on errors, report more errors
Zbigniew Jędrzejewski-Szmek [Tue, 7 May 2024 16:54:24 +0000 (18:54 +0200)] 
preset-all: continue on errors, report more errors

Firstly, if we encounter an error when iterating over the directory, gather
the error but continue. This is unlikely to happen, but if it happens, then
it doesn't seem very useful to break the preset processing at a random
point. If we can't process a unit — too bad, but since we already might
have processed some units earlier, we might as well try to process the
remaining ones.

Secondly, add missing error codes for units that are in a bad state to the
exclusion list. Those, we report them in the changes list, but consider the
whole operation a success. (-ETXTBSY and -ENOLINK were missing.)

Thirdly, add a message generator for -ENOLINK.

Fixes https://github.com/systemd/systemd/issues/21224.

6 weeks agosystemctl: do not fall back to StartUnit automatically for sleep operations 32705/head
Mike Yuan [Wed, 8 May 2024 05:41:05 +0000 (13:41 +0800)] 
systemctl: do not fall back to StartUnit automatically for sleep operations

In the majority of cases, this is caused by
sleep_supported() returning error. Hence it's
very likely that it would fail again, so
the fallback is not really useful. Instead,
honor the --force option for these verbs.

6 weeks agohibernate-util,logind: also differentiate the case of misconfigured resume
Mike Yuan [Wed, 8 May 2024 04:52:35 +0000 (12:52 +0800)] 
hibernate-util,logind: also differentiate the case of misconfigured resume

6 weeks agohibernate-util,logind: emit a clear error if the specified resume dev is missing
Mike Yuan [Wed, 8 May 2024 04:42:40 +0000 (12:42 +0800)] 
hibernate-util,logind: emit a clear error if the specified resume dev is missing

Currently, SLEEP_NOT_ENOUGH_SWAP_SPACE (ENOSPC) is returned
on all sorts of error conditions. But one important case
that's worth differentiating from that is when the resume device
is manually specified yet missing.

Closes #32644

6 weeks agoMerge pull request #32684 from YHNdnzj/pr-followups
Luca Boccassi [Tue, 7 May 2024 19:05:04 +0000 (21:05 +0200)] 
Merge pull request #32684 from YHNdnzj/pr-followups

Follow-ups for recently merged PRs

6 weeks agoMerge pull request #32681 from DaanDeMeyer/skipped-no-journal
Daan De Meyer [Tue, 7 May 2024 18:49:49 +0000 (20:49 +0200)] 
Merge pull request #32681 from DaanDeMeyer/skipped-no-journal

test: Don't keep journals for skipped tests

6 weeks agotest: Mark TEST-21-DFUZZER as slow and skip it by default 32681/head
Daan De Meyer [Tue, 7 May 2024 13:21:27 +0000 (15:21 +0200)] 
test: Mark TEST-21-DFUZZER as slow and skip it by default

The test is not really useful without sanitizers, so let's mark it
as slow and not run it by default in CI.

6 weeks agotest: Add infra to mark integration tests as slow
Daan De Meyer [Tue, 7 May 2024 13:20:44 +0000 (15:20 +0200)] 
test: Add infra to mark integration tests as slow

Some integration tests take much more time than others, let's add
a test param that can be used to configure this and integrate it
with the slow-tests meson option.

6 weeks agoci: Print a helpful link to download and view a failed test's journal
Daan De Meyer [Tue, 7 May 2024 10:24:51 +0000 (12:24 +0200)] 
ci: Print a helpful link to download and view a failed test's journal

6 weeks agorepart: correct timespec_store usage too 32684/head
Mike Yuan [Tue, 7 May 2024 12:15:57 +0000 (20:15 +0800)] 
repart: correct timespec_store usage too

6 weeks agobasic/mkdir: use timespec_store instead of _nsec for mkdir_p_root_full
Mike Yuan [Tue, 7 May 2024 12:13:55 +0000 (20:13 +0800)] 
basic/mkdir: use timespec_store instead of _nsec for mkdir_p_root_full

Follow-up for 34c3d574742e867ef97e79509e4051a82f1b7d9b

O_RDONLY is dropped when O_DIRECTORY is specified, since
it's unnecessary and even arguably confusing here, as
the dir is modified.

6 weeks agocore/service: use switch for service_enter_start
Mike Yuan [Tue, 7 May 2024 11:08:31 +0000 (19:08 +0800)] 
core/service: use switch for service_enter_start

Plus, add an assertion to ensure we're actually
using the start timestamp for the correct pid.

6 weeks agocore/execute: re-break function declaration of exec_spawn
Mike Yuan [Tue, 7 May 2024 11:02:32 +0000 (19:02 +0800)] 
core/execute: re-break function declaration of exec_spawn

Also rearrange some variable definitions. Note that
DECIMAL_STR_MAX() contains the size of NUL byte,
so +1 is dropped.

6 weeks agocore/unit: use FOREACH_STRING where appropriate
Mike Yuan [Tue, 7 May 2024 10:58:32 +0000 (18:58 +0800)] 
core/unit: use FOREACH_STRING where appropriate

Follow-up for 61628287bd6d5115d96a4bdbd70964efd7a5940e

6 weeks agoci: Optimize pull request labeler
Daan De Meyer [Tue, 7 May 2024 13:53:35 +0000 (15:53 +0200)] 
ci: Optimize pull request labeler

We keep running into rate limits, so let's optimize the number of
requests we do in the pull request labeler to hopefully fix that.

6 weeks agotest: Don't keep journals for skipped tests
Daan De Meyer [Tue, 7 May 2024 09:50:11 +0000 (11:50 +0200)] 
test: Don't keep journals for skipped tests

Let's make sure we don't save journals for tests that were skipped.

6 weeks agoMerge pull request #32683 from dtardon/fix-error-vars
Luca Boccassi [Tue, 7 May 2024 15:19:54 +0000 (17:19 +0200)] 
Merge pull request #32683 from dtardon/fix-error-vars

More mismatched error variables

6 weeks agomkosi: Update centos submodule to latest
Daan De Meyer [Tue, 7 May 2024 13:05:53 +0000 (15:05 +0200)] 
mkosi: Update centos submodule to latest

6 weeks agoclean-ipc: pass the right error variable 32683/head
David Tardon [Tue, 7 May 2024 11:56:26 +0000 (13:56 +0200)] 
clean-ipc: pass the right error variable

6 weeks agoudev-rules: pass the right error variable
David Tardon [Tue, 7 May 2024 11:55:02 +0000 (13:55 +0200)] 
udev-rules: pass the right error variable

6 weeks agohomework-quota: pass the right error variable
David Tardon [Tue, 7 May 2024 11:54:21 +0000 (13:54 +0200)] 
homework-quota: pass the right error variable

6 weeks agohomework-blob: pass the right error variable
David Tardon [Tue, 7 May 2024 11:50:55 +0000 (13:50 +0200)] 
homework-blob: pass the right error variable

6 weeks agohomework-fscrypt: pass the right error variable
David Tardon [Tue, 7 May 2024 11:49:31 +0000 (13:49 +0200)] 
homework-fscrypt: pass the right error variable

6 weeks agohomed-manager: pass the right error variable
David Tardon [Tue, 7 May 2024 11:46:32 +0000 (13:46 +0200)] 
homed-manager: pass the right error variable

6 weeks agotest-bpf-foreign-programs: pass the right error variable
David Tardon [Tue, 7 May 2024 11:43:09 +0000 (13:43 +0200)] 
test-bpf-foreign-programs: pass the right error variable

6 weeks agotest-bpf-restrict-fs: pass the right error variable
David Tardon [Tue, 7 May 2024 11:42:29 +0000 (13:42 +0200)] 
test-bpf-restrict-fs: pass the right error variable

6 weeks agotest-socket-bind: pass the right error variable
David Tardon [Tue, 7 May 2024 11:41:30 +0000 (13:41 +0200)] 
test-socket-bind: pass the right error variable

6 weeks agonspawn: pass the right error variable
David Tardon [Tue, 7 May 2024 11:40:53 +0000 (13:40 +0200)] 
nspawn: pass the right error variable

6 weeks agouserns-restrict: pass the right error variable
David Tardon [Tue, 7 May 2024 11:35:55 +0000 (13:35 +0200)] 
userns-restrict: pass the right error variable

6 weeks agocgroup-show: pass the right error variable
David Tardon [Tue, 7 May 2024 11:30:37 +0000 (13:30 +0200)] 
cgroup-show: pass the right error variable

6 weeks agomanager: pass the right error variable
David Tardon [Tue, 7 May 2024 11:29:30 +0000 (13:29 +0200)] 
manager: pass the right error variable

6 weeks agoresolved-dns-stream: pass the right error variable
David Tardon [Tue, 7 May 2024 11:23:01 +0000 (13:23 +0200)] 
resolved-dns-stream: pass the right error variable

6 weeks agoexec-invoke: pass the right error variable
David Tardon [Tue, 7 May 2024 11:19:42 +0000 (13:19 +0200)] 
exec-invoke: pass the right error variable

6 weeks agomain: pass the right error variable
David Tardon [Tue, 7 May 2024 11:18:10 +0000 (13:18 +0200)] 
main: pass the right error variable

6 weeks agobless-boot: pass the right error variable
David Tardon [Tue, 7 May 2024 11:16:30 +0000 (13:16 +0200)] 
bless-boot: pass the right error variable

6 weeks agojournal-remote-main: pass the right error variable
David Tardon [Tue, 7 May 2024 11:13:24 +0000 (13:13 +0200)] 
journal-remote-main: pass the right error variable

6 weeks agodiscover-image: pass the right error variable
David Tardon [Tue, 7 May 2024 07:07:54 +0000 (09:07 +0200)] 
discover-image: pass the right error variable

6 weeks agobuild(deps): bump pkg/debian from `50b20e2` to `7cc898c`
dependabot[bot] [Tue, 7 May 2024 09:56:09 +0000 (09:56 +0000)] 
build(deps): bump pkg/debian from `50b20e2` to `7cc898c`

Bumps pkg/debian from `50b20e2` to `7cc898c`.

---
updated-dependencies:
- dependency-name: pkg/debian
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
6 weeks agomkosi: Switch to fedora 40
Daan De Meyer [Tue, 7 May 2024 07:33:10 +0000 (09:33 +0200)] 
mkosi: Switch to fedora 40

Enable updates-testing and use the most recent mirror to make sure
we get util-linux 2.40.1 which contains a crucial fix to make sure
the serial terminal in virtual machines works properly.

6 weeks agocryptsetup: make type-specific checks for PKCS#11 private keys
Vladimir Stoiakin [Sun, 18 Feb 2024 17:40:09 +0000 (20:40 +0300)] 
cryptsetup: make type-specific checks for PKCS#11 private keys

6 weeks agoMerge pull request #32677 from keszybz/wording-fixes
Luca Boccassi [Tue, 7 May 2024 09:03:44 +0000 (11:03 +0200)] 
Merge pull request #32677 from keszybz/wording-fixes

Wording fixes

6 weeks agodocs: minor improvements to CREDENTIALS.md
Adrian Wannenmacher [Mon, 6 May 2024 18:53:02 +0000 (20:53 +0200)] 
docs: minor improvements to CREDENTIALS.md

This commit fixes two instances of a miscount. As the number is not important in either case, and seemingly subject to changes, it was removed entirely.

Another sentence was reworded to improve readability.

6 weeks agoMerge pull request #32635 from poettering/cryptenroll-no-pcrlock-conflict
Luca Boccassi [Tue, 7 May 2024 07:49:30 +0000 (09:49 +0200)] 
Merge pull request #32635 from poettering/cryptenroll-no-pcrlock-conflict

cryptenroll: don't try to use pcrlock in combination with signed PCR policy if both are available, because we don't actually support that right now

6 weeks agomanager: resolve verb/noun mixup in log message 32677/head
Zbigniew Jędrzejewski-Szmek [Wed, 24 Apr 2024 12:49:12 +0000 (14:49 +0200)] 
manager: resolve verb/noun mixup in log message

The word is used in "%s requested" so it should be a noun.

6 weeks agorepart: respect `SOURCE_DATE_EPOCH` on `mkdir_p_root`
Moritz Sanft [Mon, 29 Apr 2024 11:03:40 +0000 (13:03 +0200)] 
repart: respect `SOURCE_DATE_EPOCH` on `mkdir_p_root`

This let's systemd-repart respect the `SOURCE_DATE_EPOCH` environment
variable when creating directories in the local tree through `CopyFiles`
or `MakeDirectories`.

To do this, we pass a timestamp `ts` to `mkdir_p_root`, which it will
use to fix up `mtime` and `atime` of the directory it creates as
well as the `mtime` of the directory it creates the other directory *in*,
as the `mtime` of the latter is modified when creating a directory in it.

For the same reason, it also needs to fixup the `mtime` of the upper
directory when copying a file into it through `CopyFiles`.

If `SOURCE_DATE_EPOCH`, times are left as is. (`UTIME_OMIT`)

6 weeks agoMerge pull request #32673 from DaanDeMeyer/mkosi
Luca Boccassi [Tue, 7 May 2024 07:25:59 +0000 (09:25 +0200)] 
Merge pull request #32673 from DaanDeMeyer/mkosi

Various mkosi fixes

6 weeks agoman: reword text and fix tense in description of sd_event_source_set_io_fd
Zbigniew Jędrzejewski-Szmek [Tue, 23 Apr 2024 17:44:23 +0000 (19:44 +0200)] 
man: reword text and fix tense in description of sd_event_source_set_io_fd

Follow-up for 2fa480592d4f4334881361c5558f563e5ea4c9c3.

6 weeks agohwdb: ieee1394-unit-function: add Tascam IF-FW/DM mkII
Takashi Sakamoto [Mon, 6 May 2024 08:46:00 +0000 (17:46 +0900)] 
hwdb: ieee1394-unit-function: add Tascam IF-FW/DM mkII

Jörg M. Sigle contriutes configuration ROM image for Tascam IF-FW/DM mkII,
which includes a quirk that the values of model fields in both root and
unit directory entries are different. In device attributes, it appears
that the value of model attribute in unit device is different from the one
in node device.

This commit adds hwdb entry for the model.

6 weeks agodoc: mention that units can be masked via credentials
Luca Boccassi [Mon, 6 May 2024 19:09:35 +0000 (20:09 +0100)] 
doc: mention that units can be masked via credentials

6 weeks agoci: Disable RuntimeBuildSources= 32673/head
Daan De Meyer [Mon, 6 May 2024 20:22:50 +0000 (22:22 +0200)] 
ci: Disable RuntimeBuildSources=

We build with debuginfo, so there's no point to starting virtiofsd
to mount the sources and build directory into the VM, so let's
disable that.

6 weeks agomkosi: Install dhcp-server on OpenSUSE
Daan De Meyer [Mon, 6 May 2024 18:20:10 +0000 (20:20 +0200)] 
mkosi: Install dhcp-server on OpenSUSE

Required for systemd-networkd-tests.py.

6 weeks agomeson: Fix formatting
Daan De Meyer [Mon, 6 May 2024 16:16:09 +0000 (18:16 +0200)] 
meson: Fix formatting

6 weeks agomkosi: Install python3-psutil
Daan De Meyer [Mon, 6 May 2024 14:21:53 +0000 (16:21 +0200)] 
mkosi: Install python3-psutil

Required to run systemd-networkd-tests.py. Already a dependency of
systemd-tests on Fedora so we don't add it there as it'll be pulled
in via the prepare script.

6 weeks agomkosi: Install radvd
Daan De Meyer [Mon, 6 May 2024 13:22:46 +0000 (15:22 +0200)] 
mkosi: Install radvd

Required for networkd integration tests.

6 weeks agotpm2-util: tweak JSON condition check 32635/head
Lennart Poettering [Mon, 6 May 2024 14:12:04 +0000 (16:12 +0200)] 
tpm2-util: tweak JSON condition check

As for the other fields let's check if the actual variable we serialize
is set before serializing it.

This shouldn't make any difference, since the pubkey and the PCR mask
should always be set together or neither, but I think it's easier to
grok this way, and makes the function nicely "dumb": it serializes what
is specified, without trying to be smart by suppressng specified fields.

6 weeks agocryptenroll: determine TPM enrollment parmaeters only if we actually do TPM enrollments
Lennart Poettering [Thu, 2 May 2024 16:41:36 +0000 (18:41 +0200)] 
cryptenroll: determine TPM enrollment parmaeters only if we actually do TPM enrollments

Otherwise we'll do work (and possibly generate fatal errors) where we
really shouldn't.

6 weeks agocryptenroll: do not combine pcrlock and signed PCR policies in TPM mode
Lennart Poettering [Thu, 2 May 2024 16:34:36 +0000 (18:34 +0200)] 
cryptenroll: do not combine pcrlock and signed PCR policies in TPM mode

We currently do not support pcrlock policies and signed PCR policies in
combination. Hence, when we auto-discover both, let's disable signed PCR
policies if pcrlock is available too (simple because that covers more
ground).

Fixes: #32565
6 weeks agoMerge pull request #32664 from DaanDeMeyer/no-build
Daan De Meyer [Mon, 6 May 2024 14:01:24 +0000 (16:01 +0200)] 
Merge pull request #32664 from DaanDeMeyer/no-build

mkosi: Skip build if NO_BUILD is set

6 weeks agomkosi: Configure coredump to store coredumps in the journal. 32664/head
Daan De Meyer [Mon, 6 May 2024 12:44:43 +0000 (14:44 +0200)] 
mkosi: Configure coredump to store coredumps in the journal.

Makes it easier to debug coredumps that happen in CI as we can retrieve
them via the journal.

6 weeks agomkosi: Add missing SPDX line
Daan De Meyer [Mon, 6 May 2024 12:44:27 +0000 (14:44 +0200)] 
mkosi: Add missing SPDX line

6 weeks agomkosi: Only look for services in /usr and /etc
Daan De Meyer [Mon, 6 May 2024 12:39:28 +0000 (14:39 +0200)] 
mkosi: Only look for services in /usr and /etc

6 weeks agomkosi: Skip build if NO_BUILD is set
Daan De Meyer [Mon, 6 May 2024 12:11:13 +0000 (14:11 +0200)] 
mkosi: Skip build if NO_BUILD is set

This allows running the integration tests against the distro's packaged
version of systemd.

6 weeks agoMerge pull request #32606 from DaanDeMeyer/mkosi
Daan De Meyer [Mon, 6 May 2024 11:11:43 +0000 (13:11 +0200)] 
Merge pull request #32606 from DaanDeMeyer/mkosi

 mkosi: Enable more integration tests

6 weeks agosystemd-boot: Allow key enroll in AuditMode
Nicolas Bouchinet [Fri, 3 May 2024 09:42:09 +0000 (11:42 +0200)] 
systemd-boot: Allow key enroll in AuditMode

Since AuditMode automatically switches SetupMode on, it should be
authorized to enroll SecureBoot keys.

Signed-off-by: Nicolas Bouchinet <nicolas.bouchinet@ssi.gouv.fr>
6 weeks agomkosi: Enable more integration tests 32606/head
Daan De Meyer [Wed, 1 May 2024 07:37:55 +0000 (09:37 +0200)] 
mkosi: Enable more integration tests

6 weeks agoci: Reduce the number of integration tests we run concurrently
Daan De Meyer [Fri, 3 May 2024 08:57:22 +0000 (10:57 +0200)] 
ci: Reduce the number of integration tests we run concurrently

Since there's a bunch of CPU hungry systemd-journal-remote processes
running on the host to received the forwarded logs, by running as many
test as the VM has cores we overload the available resources. Let's leave
use the number of cores - 1 to reduce resource contention.

6 weeks agotest: Default to linux qemu firmware
Daan De Meyer [Sun, 5 May 2024 16:14:44 +0000 (18:14 +0200)] 
test: Default to linux qemu firmware

Direct kernel boot results in much faster boot times so let's use
it by default.

We disable it for tests that need to reboot because +-50% of the
time, doing a reboot when using direct kernel boot causes qemu to
hang on reboot. Until we figure that out, let's use UEFI for the
tests that need to reboot.

6 weeks agoTEST-46-HOMED: Skip barely fits test on ext4
Daan De Meyer [Mon, 6 May 2024 09:47:34 +0000 (11:47 +0200)] 
TEST-46-HOMED: Skip barely fits test on ext4

For some reason this fails on ext4 with "No space left on device".
Until we figure out why, let's skip the test on ext4 (which is reported
as ext2/ext3 by stat).

6 weeks agoTEST-46-HOMED: Only run resize tests on btrfs
Daan De Meyer [Mon, 6 May 2024 09:46:52 +0000 (11:46 +0200)] 
TEST-46-HOMED: Only run resize tests on btrfs

Other filesystems do not support online shrinking.

6 weeks agoTEST-53-ISSUE-16347: Add rtc configuration for qemu
Daan De Meyer [Sun, 5 May 2024 20:17:38 +0000 (22:17 +0200)] 
TEST-53-ISSUE-16347: Add rtc configuration for qemu

6 weeks agoTEST-55-OOMD: Configure init.scope credential in mkosi image
Daan De Meyer [Sun, 5 May 2024 08:46:01 +0000 (10:46 +0200)] 
TEST-55-OOMD: Configure init.scope credential in mkosi image

6 weeks agoTEST-55-OOMD: Skip on opensuse
Daan De Meyer [Sun, 5 May 2024 08:39:32 +0000 (10:39 +0200)] 
TEST-55-OOMD: Skip on opensuse

opensuse does not have the stress tool packaged.

6 weeks agoTEST-55-OOMD: swapoff before adding new swapfile
Richard Maw [Mon, 19 Feb 2024 17:28:00 +0000 (17:28 +0000)] 
TEST-55-OOMD: swapoff before adding new swapfile

When running test images built with read-only /usr
a swap partition is likely so needs to be turned off first.