]> git.ipfire.org Git - thirdparty/systemd.git/log
thirdparty/systemd.git
20 months agoman: similar → similarly
Zbigniew Jędrzejewski-Szmek [Tue, 23 Aug 2022 10:12:28 +0000 (12:12 +0200)] 
man: similar → similarly

Something *is* similar
Something *works* similarly
Something does something, similarly to how something else does something

See https://sites.ulethbridge.ca/roussel/2017/11/29/similar-and-similarly-are-they-similar/
for a clear explanation.

21 months agotake jointly mounted controllers into account when migrate cgroups, otherwise "CGroup...
jiangchuangang [Mon, 22 Aug 2022 03:13:51 +0000 (11:13 +0800)] 
take jointly mounted controllers into account when migrate cgroups, otherwise "CGroupMask done" seems to be redundant.

21 months agoMerge pull request #24379 from mrc0mmand/llvm-15 23803/head
Luca Boccassi [Sat, 20 Aug 2022 22:38:52 +0000 (23:38 +0100)] 
Merge pull request #24379 from mrc0mmand/llvm-15

ci: build with clang-15; drop clang-12

21 months agomkosi: Install tests in final image
Daan De Meyer [Tue, 16 Aug 2022 10:50:29 +0000 (12:50 +0200)] 
mkosi: Install tests in final image

21 months agohashmap: use assert_se() to make clang happy 24379/head
Frantisek Sumsal [Sat, 20 Aug 2022 19:57:18 +0000 (21:57 +0200)] 
hashmap: use assert_se() to make clang happy

Otherwise it complains about a set but unused variable:

```
../src/basic/hashmap.c:1070:48: error: variable 'n_rehashed' set but not used [-Werror,-Wunused-but-set-variable]
        unsigned old_n_buckets, new_n_buckets, n_rehashed, new_n_entries;
                                                       ^
                                                       1 error generated.
```

21 months agocoredump: drop an unused variable
Frantisek Sumsal [Sat, 20 Aug 2022 19:04:24 +0000 (21:04 +0200)] 
coredump: drop an unused variable

21 months agonetwork: drop an unused variable
Frantisek Sumsal [Sat, 20 Aug 2022 19:00:14 +0000 (21:00 +0200)] 
network: drop an unused variable

21 months agomachine: drop an unused variable
Frantisek Sumsal [Sat, 20 Aug 2022 18:45:27 +0000 (20:45 +0200)] 
machine: drop an unused variable

21 months agosd-journal: drop an unused variable
Frantisek Sumsal [Sat, 20 Aug 2022 18:38:17 +0000 (20:38 +0200)] 
sd-journal: drop an unused variable

21 months agoci: build with clang-15; drop clang-12
Frantisek Sumsal [Sat, 20 Aug 2022 18:11:32 +0000 (20:11 +0200)] 
ci: build with clang-15; drop clang-12

21 months agoMerge pull request #24378 from yuwata/test-network-issue-24377
Yu Watanabe [Sat, 20 Aug 2022 13:24:53 +0000 (22:24 +0900)] 
Merge pull request #24378 from yuwata/test-network-issue-24377

test-network: add test case for issue #24377

21 months agotest-network: add test for issue #24377 24378/head
Yu Watanabe [Sat, 20 Aug 2022 11:36:12 +0000 (20:36 +0900)] 
test-network: add test for issue #24377

The issue has been already fixed by
b05e52000b4eee764b383cc3031da0a3739e996e (PR #24020).

21 months agotest-network: add helper functions for reading logs of networkd
Yu Watanabe [Sat, 20 Aug 2022 11:35:18 +0000 (20:35 +0900)] 
test-network: add helper functions for reading logs of networkd

21 months agoMerge pull request #24356 from keszybz/sd-netlink-api
Yu Watanabe [Sat, 20 Aug 2022 10:11:02 +0000 (19:11 +0900)] 
Merge pull request #24356 from keszybz/sd-netlink-api

Small improvements to the sd-netlink api

21 months agohwdb: Add Avita Liber NS13A2 (#24376)
Aryan singh [Sat, 20 Aug 2022 10:02:54 +0000 (15:32 +0530)] 
hwdb: Add Avita Liber NS13A2 (#24376)

This fixes the discrepancies in the coordinate ranges for the touchpad, touchpad in this device(NS13A2) is generic and the same one is used in most models.

21 months agomeasure: store bank name in PcrState structure
Lennart Poettering [Wed, 17 Aug 2022 15:10:53 +0000 (17:10 +0200)] 
measure: store bank name in PcrState structure

Let's not query/strdup()ascii_strlower() it all the time, but just cache
it once and use it.

21 months agotest: wait for whole block device instead of partition
Yu Watanabe [Fri, 19 Aug 2022 22:09:21 +0000 (07:09 +0900)] 
test: wait for whole block device instead of partition

In 'udevadm lock' the device /dev/loopX is locked instead of
/dev/loopXp1. Hence, 'udevadm wait' should wait for /dev/loopX.

For some reasons, the kernel sometimes does not emit uevent for
partitions, and 'udevadm wait' for partitions may fail.

Fixes #24360.

21 months agotest: correctly process multiline strings in $KERNEL_APPEND
Frantisek Sumsal [Fri, 19 Aug 2022 14:30:24 +0000 (16:30 +0200)] 
test: correctly process multiline strings in $KERNEL_APPEND

Some tests (like TEST-02) set a multiline string to $KERNEL_APPEND
(which is a valid thing to do), unfortunately we'd use only the first
line of it and throw the rest away, e.g:

```
$ printf "%s" "$x"
hello

this is a multiline

kernel command line
$ read -ra out <<< "$x"
$ printf "%s" "${out[@]}"
hello
```

Let's use readarray/mapfile instead to avoid this:

```
$ readarray out <<< "$x"
$ printf "%s" "${out[@]}"
hello

this is a multiline

kernel command line

```

21 months agomeasure: add json output
Lennart Poettering [Fri, 12 Aug 2022 13:23:44 +0000 (15:23 +0200)] 
measure: add json output

21 months agoMerge pull request #24368 from poettering/tpm2-json-pcr-array-rework
Lennart Poettering [Fri, 19 Aug 2022 19:51:47 +0000 (21:51 +0200)] 
Merge pull request #24368 from poettering/tpm2-json-pcr-array-rework

tpm2: add helpers for building/parsing JSON arrays of PCR indexes

21 months agoMerge pull request #24350 from DaanDeMeyer/docs-pkexec
Luca Boccassi [Fri, 19 Aug 2022 19:43:50 +0000 (20:43 +0100)] 
Merge pull request #24350 from DaanDeMeyer/docs-pkexec

docs: Recommend pkexec over using an askpass program with sudo

21 months agotpm2-util: introduce tpm2_parse_pcr_argument() helper
Lennart Poettering [Wed, 17 Aug 2022 15:00:27 +0000 (17:00 +0200)] 
tpm2-util: introduce tpm2_parse_pcr_argument() helper

Add a new tpm2_parse_pcr_argument() helper that unifies how we merge PCR
masks in a single function, we can use all over the place. Previously we
had basically the same code for this at 4 places.

21 months agorules: import previous SYSTEMD_READY state for suspended DM devices and skip other...
Michal Sekletar [Tue, 2 Aug 2022 10:38:30 +0000 (12:38 +0200)] 
rules: import previous SYSTEMD_READY state for suspended DM devices and skip other rules

We can't get any FS meta-data from a suspended device. Hence defer
making any plugged/unplugged decisions, i.e. we just import whatever was
previous state and skip processing all other rules.

Thanks Lennart Poettering <lennart@poettering.net> for suggesting this
solution.

21 months agotpm2-util: expose more hash algorithms
Lennart Poettering [Wed, 17 Aug 2022 15:30:29 +0000 (17:30 +0200)] 
tpm2-util: expose more hash algorithms

swtpm supports them, hence maybe support them in our codebase, too

21 months agotpm2-util: add TPM2_PCR_MASK_VALID() helper
Lennart Poettering [Wed, 17 Aug 2022 14:49:14 +0000 (16:49 +0200)] 
tpm2-util: add TPM2_PCR_MASK_VALID() helper

21 months agosd-netlink: use SD_EVENT_ONESHOT for a time-based callback 24356/head
Zbigniew Jędrzejewski-Szmek [Fri, 19 Aug 2022 15:43:56 +0000 (17:43 +0200)] 
sd-netlink: use SD_EVENT_ONESHOT for a time-based callback

21 months agoTEST-70-TPM2: use "truncate" to generate disk image
Lennart Poettering [Fri, 19 Aug 2022 09:19:59 +0000 (11:19 +0200)] 
TEST-70-TPM2: use "truncate" to generate disk image

Let'se operate on a sparse file here, to reuduce memory use.

21 months agoMerge pull request #24362 from poettering/sha256-tweaks
Lennart Poettering [Fri, 19 Aug 2022 14:33:35 +0000 (16:33 +0200)] 
Merge pull request #24362 from poettering/sha256-tweaks

some tweaks to the sha256 implementation

21 months agotpm2-util: also add helper for parsing PCR arrays 24368/head
Lennart Poettering [Fri, 19 Aug 2022 14:16:03 +0000 (16:16 +0200)] 
tpm2-util: also add helper for parsing PCR arrays

21 months agotpm2-util: split out helpers which format a PCR mask as a JSON array
Lennart Poettering [Fri, 19 Aug 2022 14:09:51 +0000 (16:09 +0200)] 
tpm2-util: split out helpers which format a PCR mask as a JSON array

This makes the code easier to read, and is something we can reuse later
on.

21 months agotree-wide: pass proper values (not bools) to sd_event_source_set_enabled()
Lennart Poettering [Fri, 19 Aug 2022 10:40:24 +0000 (12:40 +0200)] 
tree-wide: pass proper values (not bools) to sd_event_source_set_enabled()

21 months agoMerge pull request #23653 from aafeijoo-suse/ask-for-recovery-key
Lennart Poettering [Fri, 19 Aug 2022 12:55:54 +0000 (14:55 +0200)] 
Merge pull request #23653 from aafeijoo-suse/ask-for-recovery-key

cryptsetup: improve password prompt text

21 months agoAdd --efi-boot-option-description argument to bootctl to control the name of the...
Rene Hollander [Sat, 13 Aug 2022 19:38:02 +0000 (21:38 +0200)] 
Add --efi-boot-option-description argument to bootctl to control the name of the boot
entry.

By default an entry named "Linux Boot Manager" is created (which is the
previous behavior). With the flag the name of the entry can be
controlled, which is useful when installing systemd-boot to multiple ESP
partitions and having uniquely named entries.

Fixes #17044.

21 months agoTEST-70-TPM2: actually enable systemd-measure test
Lennart Poettering [Fri, 19 Aug 2022 09:23:45 +0000 (11:23 +0200)] 
TEST-70-TPM2: actually enable systemd-measure test

A bit emberassing that this never worked.

21 months agocryptsetup: make sure all token-based codepaths are effected by SYSTEMD_CRYPTSETUP_US...
Lennart Poettering [Fri, 19 Aug 2022 09:33:41 +0000 (11:33 +0200)] 
cryptsetup: make sure all token-based codepaths are effected by SYSTEMD_CRYPTSETUP_USE_TOKEN_MODULE env var

Previously the env var was only checked when conditionalizing use of our
own libcryptsetup loadable token modules.  But let's also use it for any
other kind of token module, including possible internal ones by
libcryptsetup.

21 months agocryptsetup: use right internal helper when checking whether to use tokens
Lennart Poettering [Fri, 19 Aug 2022 09:32:07 +0000 (11:32 +0200)] 
cryptsetup: use right internal helper when checking whether to use tokens

The other codepaths get this right, the TPM2 one currently does not. Fix
that.

21 months agosha256: add sha256_direct()/SHA256_DIRECT() helpers 24362/head
Lennart Poettering [Wed, 17 Aug 2022 09:32:38 +0000 (11:32 +0200)] 
sha256: add sha256_direct()/SHA256_DIRECT() helpers

21 months agorandom-seed: use SHA256_DIGEST_SIZE
Lennart Poettering [Wed, 17 Aug 2022 09:31:45 +0000 (11:31 +0200)] 
random-seed: use SHA256_DIGEST_SIZE

21 months agosha256: change digest buffer type to uint8_t[]
Lennart Poettering [Wed, 17 Aug 2022 09:24:24 +0000 (11:24 +0200)] 
sha256: change digest buffer type to uint8_t[]

This way we can specify a size with "static". All users use uint8_t
already, hence this comes at no price.

21 months agosd-netlink: group message cons methods together
Zbigniew Jędrzejewski-Szmek [Fri, 19 Aug 2022 08:39:37 +0000 (10:39 +0200)] 
sd-netlink: group message cons methods together

21 months agosd-netlink: simplify error code retention
Zbigniew Jędrzejewski-Szmek [Thu, 18 Aug 2022 13:06:00 +0000 (15:06 +0200)] 
sd-netlink: simplify error code retention

21 months agosd-netlink: delete sd_netlink_new_from_fd()
Zbigniew Jędrzejewski-Szmek [Thu, 18 Aug 2022 12:24:14 +0000 (14:24 +0200)] 
sd-netlink: delete sd_netlink_new_from_fd()

It was somewhat similar to sd_netlink_open_fd(), and unused.

21 months agosd-netlink: rename sd_netlink_message_request_dump to sd_netlink_message_set_request_dump
Zbigniew Jędrzejewski-Szmek [Thu, 18 Aug 2022 09:08:44 +0000 (11:08 +0200)] 
sd-netlink: rename sd_netlink_message_request_dump to sd_netlink_message_set_request_dump

21 months agosd-netlink: constify parameters in public api and adjust pointer formatting
Zbigniew Jędrzejewski-Szmek [Thu, 18 Aug 2022 09:04:16 +0000 (11:04 +0200)] 
sd-netlink: constify parameters in public api and adjust pointer formatting

In the internal api, 'char* const*' is used instead of 'const char* const*'
because otherwise we'd get a mismatch with the various strv apis.

21 months agosd-netlink: prefix output param names with 'ret'
Zbigniew Jędrzejewski-Szmek [Thu, 18 Aug 2022 08:50:41 +0000 (10:50 +0200)] 
sd-netlink: prefix output param names with 'ret'

In some places, the code was already using that, so only the header file needed
to be adjusted.

We use 'nl' for the sd_netlink* when it's passed in as input. I was considering
renaming to 'netlink', but that'd add a lot of verbosity and also we have
'sd_journal *j' in sd-journal, so the short name matches in style.

21 months agoDrop the limit on number of inodes for /dev
Franck Bui [Mon, 6 Dec 2021 14:00:12 +0000 (15:00 +0100)] 
Drop the limit on number of inodes for /dev

Follow-up for 4c733d3046942984c5f73b40c3af39cc218c103f.

Finding a suitable limit that would fit any use cases out there is pretty hard
and since /dev is only writeable by root anyway, let's simply drop the limit.

21 months agoupdate TODO 23653/head
Antonio Alvarez Feijoo [Fri, 19 Aug 2022 07:02:17 +0000 (09:02 +0200)] 
update TODO

21 months agocryptsetup: improve password prompt text
Antonio Alvarez Feijoo [Fri, 19 Aug 2022 07:01:56 +0000 (09:01 +0200)] 
cryptsetup: improve password prompt text

Instead of always asking for passphrase, if the device has LUKS2 header check:
- If only regular passphrases are registered, ask for passphrase.
- If only recovery keys are registered, ask for recovery key.
- If both regular passphrases and recovery keys are registered, ask for
passphrase or recovery key.

21 months agoMerge pull request #24353 from alpernebbi/hwdb-cros-ec-accel-base
Yu Watanabe [Fri, 19 Aug 2022 05:39:47 +0000 (14:39 +0900)] 
Merge pull request #24353 from alpernebbi/hwdb-cros-ec-accel-base

udev: hwdb: Add orientation quirk for base-mounted accelerometers on Chromebooks

Reviewed-by: Alexandru Stan <amstan@chromium.org>
Reviewed-by: Gwendal Grignou <gwendal@chromium.org>
21 months agoMerge pull request #24348 from yuwata/network-ipv4acd-renew
Yu Watanabe [Fri, 19 Aug 2022 05:27:05 +0000 (14:27 +0900)] 
Merge pull request #24348 from yuwata/network-ipv4acd-renew

network: fix IPv4ACD issues on DHCP renewal or reconfiguring interface

21 months agoudev: hwdb: Add rules to match cros-ec-accel by 'location' sysfs file 24353/head
Alper Nebi Yasak [Thu, 18 Aug 2022 16:56:13 +0000 (19:56 +0300)] 
udev: hwdb: Add rules to match cros-ec-accel by 'location' sysfs file

The cros-ec-accel devices report their mounting location by the 'label'
sysfs file only since Linux v6.0. With earlier kernels, a nonstandard
'location' file reports this, but slightly differently (lid instead of
display) [1].

Add udev rules to import the correct hwdb entries based on this
'location' file for cros-ec-accel devices, so that the base-mounted
accel matrix has the correct value for older kernels as well.

[1] https://kernel.org/doc/Documentation/ABI/testing/sysfs-bus-iio-cros-ec

21 months agohwdb: Add Chromebook accel orientation quirks based on sysfs label
Alper Nebi Yasak [Thu, 18 Aug 2022 15:24:05 +0000 (18:24 +0300)] 
hwdb: Add Chromebook accel orientation quirks based on sysfs label

The base-mounted accelerometer on Chromebooks return values same as the
display when the lid angle is 180 degrees, instead of when the lid is
closed. To match userspace expectations we must further rotate the
existing accelerometer mounting matrix by 180 degrees around the X axis:

    [[-1,  0,  0],    [[ 1,  0,  0],    [[-1,  0,  0],
     [ 0, -1,  0],  X  [ 0, -1,  0],  =  [ 0,  1,  0],
     [ 0,  0, -1]]     [ 0,  0, -1]]     [ 0,  0,  1]]

A previous commit lets us distinguish between the two cros-ec-accel
devices on these boards by their 'label' sysfs file. Add hwdb entries
that make base-mounted accelerometers use this correct matrix, and
display-mounted ones use the existing one.

Note that the cros-ec-accel drivers use 'label' only since Linux v6.0.
The old match strings are not removed to support older kernels, even
though they are only correct for the display-mounted sensor.

21 months agoudev: hwdb: Match iio sensors based on their label
Alper Nebi Yasak [Thu, 18 Aug 2022 14:27:22 +0000 (17:27 +0300)] 
udev: hwdb: Match iio sensors based on their label

The IIO subsystem exposes a 'label' sysfs file to help userspace better
identify its devices [1]. Standardized labels include the sensor type
along with its location, including 'accel-base' and 'accel-display'.

Most Chrome OS boards have two accelerometers that are indistinguishable
except for this label (or a 'location' sysfs file before Linux v6.0),
and need different mounting matrix corrections based on their location.

Add a udev rule that matches hwdb entries using this label, so we can
correct both accelerometers on these devices with hwdb entries. The
existing rules and hwdb entries are not modified to keep potential
out-of-tree entries working, but new entries in this form will override
existing ones. Also add currently standardized labels to parse-hwdb.py.

[1] https://www.kernel.org/doc/Documentation/ABI/testing/sysfs-bus-iio

21 months agoMerge pull request #24054 from keszybz/initrd-no-reload
Frantisek Sumsal [Thu, 18 Aug 2022 13:15:14 +0000 (13:15 +0000)] 
Merge pull request #24054 from keszybz/initrd-no-reload

Don't do daemon-reload in the initrd

21 months agodocs: Add guide to debug binaries in vscode that need to run as root 24350/head
Daan De Meyer [Thu, 18 Aug 2022 12:38:19 +0000 (14:38 +0200)] 
docs: Add guide to debug binaries in vscode that need to run as root

21 months agodocs: Recommend pkexec over using an askpass program with sudo
Daan De Meyer [Thu, 18 Aug 2022 12:16:08 +0000 (14:16 +0200)] 
docs: Recommend pkexec over using an askpass program with sudo

Less configuration required than downloading and configuring an
askpass program.

21 months agobash-completion: autocomplete cgroup names in systemd-cgtop
Luca Boccassi [Tue, 16 Aug 2022 22:04:40 +0000 (23:04 +0100)] 
bash-completion: autocomplete cgroup names in systemd-cgtop

21 months agoinitrd-parse-etc: override argv[0] to avoid dracut issue 24054/head
Zbigniew Jędrzejewski-Szmek [Sat, 13 Aug 2022 13:51:14 +0000 (15:51 +0200)] 
initrd-parse-etc: override argv[0] to avoid dracut issue

Quoting https://github.com/systemd/systemd/pull/24054#issuecomment-1210501631:
> this would need a patch in dracut, specifically adding the
> systemd-sysroot-fstab-check to the list of installed stuff:
> https://github.com/dracutdevs/dracut/blob/fe8fa2b0cadbb33e27c8dd8b5851548dcd65835c/modules.d/00systemd/module-setup.sh#L47.
>
> I could do this manually in the CI (and I guess I'd have to do it anyway even
> if the patch lands in upstream, since it won't be available in C8S), but it
> should get there first before merging this PR, otherwise it's going to break
> Rawhide.

21 months agotest-network: add tests for IPv4ACD and renewing DHCP address 24348/head
Yu Watanabe [Thu, 18 Aug 2022 06:16:27 +0000 (15:16 +0900)] 
test-network: add tests for IPv4ACD and renewing DHCP address

21 months agotest-network: drop unused .network file
Yu Watanabe [Thu, 18 Aug 2022 05:49:42 +0000 (14:49 +0900)] 
test-network: drop unused .network file

21 months agonetwork: unref existing sd_ipv4acd object when not necessary
Yu Watanabe [Thu, 18 Aug 2022 06:39:22 +0000 (15:39 +0900)] 
network: unref existing sd_ipv4acd object when not necessary

On reconfiguring an interface, the new setting may not enable IPv4ACD
for an existing address anymore. Hence, we need to unref it. Otherwise,
newly requested addresses may never be ready for (re-)configuring.

21 months agonetwork: update setting for IPv4ACD or IPv6DAD in existing Address objects
Yu Watanabe [Thu, 18 Aug 2022 05:32:47 +0000 (14:32 +0900)] 
network: update setting for IPv4ACD or IPv6DAD in existing Address objects

Otherwise, if the setting is changed on reconfigure, the new setting
is not applied to the existing addresses.

21 months agonetwork: fix DHCPv4 address renewal with IPv4ACD
Yu Watanabe [Thu, 18 Aug 2022 05:15:23 +0000 (14:15 +0900)] 
network: fix DHCPv4 address renewal with IPv4ACD

Previously, when a DHCP address is renewed and if the IPv4ACD for the
address is enabled, the address will never drop the probing flag, thus
the lifetime of the address will never be updated.

This drops NETWORK_CONFIG_STATE_PROBING, and the IPv4ACD status is
managed another bit, Address.acd_bound. And, the flag is updated only
when the IPv4ACD announced the address or detects conflict.

21 months agotest: expect libdevmapper failure in TEST-50-DISSECT
Luca Boccassi [Wed, 17 Aug 2022 18:43:01 +0000 (19:43 +0100)] 
test: expect libdevmapper failure in TEST-50-DISSECT

libdevmapper/device mapper driver can return semi-random failures when
opening verity devices, and we have fallback code to deal with it.
But the test was not expecting the fallback path, so it became unreliable.

Fixes https://github.com/systemd/systemd/issues/23866

21 months agosd-messages: rename newly added constants
Zbigniew Jędrzejewski-Szmek [Wed, 17 Aug 2022 11:09:49 +0000 (13:09 +0200)] 
sd-messages: rename newly added constants

This hasn't been through a release yet, so we are free to change the name.

Closes #24270.
C.f. https://github.com/systemd/systemd/commit/907506695549c689710cd267583f6fd439711dcf#r80812414

21 months agosd-messages: make the table wider
Zbigniew Jędrzejewski-Szmek [Wed, 17 Aug 2022 11:04:49 +0000 (13:04 +0200)] 
sd-messages: make the table wider

For whatever reason, nowadays we add entries that are quite long, so
we were using line continuation for most of new entries. This is hard
to read and edit. So let's just make the table wide enough to accomodate
all the names without line splits.

21 months agotpm2-util: more structure initialization simplifications
Lennart Poettering [Wed, 17 Aug 2022 07:49:41 +0000 (09:49 +0200)] 
tpm2-util: more structure initialization simplifications

21 months agorandom-util: drop unnecessary header file
Lennart Poettering [Wed, 17 Aug 2022 07:45:19 +0000 (09:45 +0200)] 
random-util: drop unnecessary header file

(And some minor reindents)

21 months agoMerge pull request #24250 from yuwata/backlight-multiple-graphics-cards
Yu Watanabe [Wed, 17 Aug 2022 12:58:59 +0000 (21:58 +0900)] 
Merge pull request #24250 from yuwata/backlight-multiple-graphics-cards

backlight: support multiple graphics cards system

21 months agofs-util: make touch() an inline function
Lennart Poettering [Wed, 17 Aug 2022 07:45:04 +0000 (09:45 +0200)] 
fs-util: make touch() an inline function

21 months agotpm2-util: credit TPM2 RNG entropy only once per boot
Lennart Poettering [Wed, 17 Aug 2022 07:51:35 +0000 (09:51 +0200)] 
tpm2-util: credit TPM2 RNG entropy only once per boot

Acquiring random data from the TPM is not precisely quick, let's speed
things up by doing this at most once per boot. For that, let's maintain
a flag file in /run/.

21 months agoudev: rename various validate() -> should_reload() for builtin commands
Yu Watanabe [Wed, 17 Aug 2022 05:44:27 +0000 (14:44 +0900)] 
udev: rename various validate() -> should_reload() for builtin commands

Previously, true by validate() means several configs are outdated and we
need to reload configs. That's not intuitive for me. Let's rename the
functions.

21 months agoMerge pull request #24333 from yuwata/sysctl
Yu Watanabe [Wed, 17 Aug 2022 12:56:15 +0000 (21:56 +0900)] 
Merge pull request #24333 from yuwata/sysctl

sysctl: improve performance for applying glob pattern

21 months agotest-network: add/update module check
Yu Watanabe [Tue, 16 Aug 2022 17:43:16 +0000 (02:43 +0900)] 
test-network: add/update module check

For https://github.com/systemd/systemd-centos-ci/pull/517.

21 months agotest: add tests for glob sysctl pattern 24333/head
Yu Watanabe [Tue, 16 Aug 2022 20:22:48 +0000 (05:22 +0900)] 
test: add tests for glob sysctl pattern

21 months agotest: use assertions in sysctl tests
Yu Watanabe [Tue, 16 Aug 2022 20:33:18 +0000 (05:33 +0900)] 
test: use assertions in sysctl tests

21 months agotest: do not use sysctl.d to store test conf
Yu Watanabe [Tue, 16 Aug 2022 20:20:27 +0000 (05:20 +0900)] 
test: do not use sysctl.d to store test conf

Otherwise, late invocations of systemd-sysctl, especially through udev
rules may fail.

21 months agosysctl: apply prefix before calling glob()
Yu Watanabe [Tue, 16 Aug 2022 19:54:06 +0000 (04:54 +0900)] 
sysctl: apply prefix before calling glob()

Otherwise, if there exist million of network interfaces,
calling glob() for network properties takes much time.

Fixes #24031.

21 months agopath-util: introduce path_glob_can_match()
Yu Watanabe [Tue, 16 Aug 2022 21:43:37 +0000 (06:43 +0900)] 
path-util: introduce path_glob_can_match()

21 months agosysctl: split out code for applying glob option
Yu Watanabe [Tue, 16 Aug 2022 19:10:30 +0000 (04:10 +0900)] 
sysctl: split out code for applying glob option

21 months agosysctl: drop /proc/sys/ in prefix
Yu Watanabe [Tue, 16 Aug 2022 18:11:00 +0000 (03:11 +0900)] 
sysctl: drop /proc/sys/ in prefix

21 months agosysctl: use ordered_hashmap_ensure_put()
Yu Watanabe [Wed, 17 Aug 2022 05:29:26 +0000 (14:29 +0900)] 
sysctl: use ordered_hashmap_ensure_put()

21 months agobacklight: fix issue on multiple graphics cards system 24250/head
Yu Watanabe [Mon, 8 Aug 2022 17:02:02 +0000 (02:02 +0900)] 
backlight: fix issue on multiple graphics cards system

If a system has multiple graphics cards, then we cannot associate
platform backlight devices to backlight devices under PCI bus.

Previously, in such case, vaidate_device() for a raw backlight device
might erroneously detect a platform device and return false. So, users
could not save/load backlight level.

This makes validate_device() give up to associate platform devices on
non-PCI bus with raw backlight devices. That may cause unwanted
backlight level save or restore by systemd-backlight@.service, but users
can workaround that by masking specific instances of the service.

Closes #24223.

21 months agobacklight: filter out unnecessary backlight devices by device enumerator
Yu Watanabe [Mon, 8 Aug 2022 14:25:34 +0000 (23:25 +0900)] 
backlight: filter out unnecessary backlight devices by device enumerator

21 months agobacklight: add/update several logs for validating backlight devices
Yu Watanabe [Mon, 8 Aug 2022 11:57:14 +0000 (20:57 +0900)] 
backlight: add/update several logs for validating backlight devices

21 months agorun: simplification
Zbigniew Jędrzejewski-Szmek [Tue, 16 Aug 2022 16:41:00 +0000 (18:41 +0200)] 
run: simplification

21 months agomanager: allow assignment of properties on target/swap/device units
Zbigniew Jędrzejewski-Szmek [Tue, 16 Aug 2022 16:07:30 +0000 (18:07 +0200)] 
manager: allow assignment of properties on target/swap/device units

E.g. Documentation or Markers could apply to any unit type. This already worked
partially, because a direct dbus call could be made:

After rebuild with the patch, but before the manager has been restarted:
$ build/systemctl --user set-property dev-zram0.swap Markers=+needs-restart
$ build/systemctl --user show -p Markers dev-zram0.swap
Markers=needs-restart

I noticed that that the rpm unit restart helper was throwing errors for target
units. We should just let the Markers be set for those too, even if it doesn't
do anything in the end. This way we don't need to special-case by unit type.

21 months agoMerge pull request #24301 from yuwata/network-tuntap
Luca Boccassi [Tue, 16 Aug 2022 22:06:16 +0000 (23:06 +0100)] 
Merge pull request #24301 from yuwata/network-tuntap

network/tuntap: introduce KeepFileDescriptor= setting

21 months agotest-network: add tests for MTUBytes= for bridge master and ports
Yu Watanabe [Tue, 16 Aug 2022 16:59:56 +0000 (01:59 +0900)] 
test-network: add tests for MTUBytes= for bridge master and ports

Prompted by #24311.

21 months agoman: Correct information on sysext masking
Kai Lueke [Tue, 16 Aug 2022 10:29:12 +0000 (12:29 +0200)] 
man: Correct information on sysext masking

While I had tested that a symlink to /dev/null works to "mask" a sysext
I must have gotten something wrong and thus the instructions in
519c2f0d6b343d140f7e08e3eb0f46708c023b4a don't work. What works,
at least at the moment, is to instead have an empty directory with the
extension name under /etc/extensions/.
Correct the info in the man page and add a test for it.

21 months agoci(lint): add shell linter - Differential ShellCheck
Jan Macku [Tue, 16 Aug 2022 12:34:49 +0000 (14:34 +0200)] 
ci(lint): add shell linter - Differential ShellCheck

It performs differential ShellCheck scans and report results directly in
pull request.

documentation:
https://github.com/redhat-plumbers-in-action/differential-shellcheck

21 months agotest-network: add tests for KeepCarrier= for tuntap interfaces 24301/head
Yu Watanabe [Fri, 12 Aug 2022 22:46:47 +0000 (07:46 +0900)] 
test-network: add tests for KeepCarrier= for tuntap interfaces

21 months agonetwork/tuntap: save tun or tap file descriptor in fd store
Yu Watanabe [Sat, 13 Aug 2022 08:18:55 +0000 (17:18 +0900)] 
network/tuntap: save tun or tap file descriptor in fd store

21 months agonetwork/tuntap: introduce KeepCarrier= setting
Yu Watanabe [Fri, 12 Aug 2022 22:45:49 +0000 (07:45 +0900)] 
network/tuntap: introduce KeepCarrier= setting

Closes #24267.

21 months agosysusers: add fsync for passwd (#24324)
Avram Lubkin [Tue, 16 Aug 2022 12:51:21 +0000 (08:51 -0400)] 
sysusers: add fsync for passwd (#24324)

https://github.com/systemd/systemd/pull/6636 added `fsync()` when
temporary shadow, group, and gshadow files are created, but it was
not added for passwd. As far as I can tell, this seems to have been
an oversight. I'm seeing real world issues where a blank /etc/passwd
file is being created if a machine loses power early in the boot process.

21 months agonetwork/tuntap: code cleanups
Yu Watanabe [Fri, 12 Aug 2022 22:35:21 +0000 (07:35 +0900)] 
network/tuntap: code cleanups

- merge unnecessarily split functions,
- drop unnecessary initializations,
- tighten variable scopes,
- introduce TUNTAP() helper function.

21 months agologin: use helper functions for fd store
Yu Watanabe [Mon, 15 Aug 2022 11:05:32 +0000 (20:05 +0900)] 
login: use helper functions for fd store

21 months agodaemon-util: introduce several helper functions for fd store
Yu Watanabe [Mon, 15 Aug 2022 11:05:21 +0000 (20:05 +0900)] 
daemon-util: introduce several helper functions for fd store

21 months agotpm-util: use trial session where appropriate
Lennart Poettering [Mon, 15 Aug 2022 09:44:52 +0000 (11:44 +0200)] 
tpm-util: use trial session where appropriate

TPM2 knows two types of policy sessions: "real" ones and "trial" ones. The
latter allow you to calculate a policy hash without this enforcing any
policy, which the former do. Typically you want to use the "trial" ones
when enrolling, and you have to use the "real" ones for unlocking. So
far we used "real" ones for both cases. Which works fine – as long as
the policy put together matches the current reality (e.g. the PCR values
included in the policy are the ones currently in place in the TPM).

Let's switch to using trial sessions for enrolling. First of all this is
preparation for later work to implement further policy extensions (for
example, policies binding to literally specified PCR values, instead of
the once currently measured). But from my perspective more importantly
it actually is cleaner, as it communicates more clearly what we are
actually doing here.

No user-visible change in behaviour.

21 months agomkosi: Update to latest commit
Daan De Meyer [Mon, 15 Aug 2022 12:31:13 +0000 (14:31 +0200)] 
mkosi: Update to latest commit

Introduces a more reliable mirror for Arch which should reduce the
number of mkosi Arch CI failures due to unreliable mirror selection.