Yu Watanabe [Sun, 22 Feb 2026 16:01:45 +0000 (01:01 +0900)]
udev: drop home-grown udev-ctrl socket
Nowadays, varlink is used to control systemd-udevd. Let's drop the
legacy socket.
Note, the existence of /run/udev/control socket is widely used in both
our code and external projects. Also, the dependency to
systemd-udevd-control.socket is widely used in many projects.
Hence, we need to create a symlink to the socket file and .socket unit
file.
Removed note clarifying that portable services are only for system services and not user services, and changed comparisons to "system services" with just "services". With newer systemd versions, `systemd-portabled` can be run as a user service.
The purpose of this header was to provide MIT-0 sources that can be copied
and pasted liberally. Including an LGPL-2.1+ header from it deafeats its
purpose. Make it self-standing again.
udev-util: bound leading whitespace skip in udev_replace_whitespace (#42757)
udev_replace_whitespace() is documented to read at most 'len' bytes from
'str':
- the strspn() skip of leading whitespace stops at a non-space byte or
NUL, not at 'len'
- ata_id passes the space padded, non-NUL-terminated ATA IDENTIFY
model/serial/fw fields
- an all-blank field reads off the end of the 512-byte hd_driveid stack
struct
Capped the skip to 'len'; existing outputs are unchanged. Added a
regression test.
discover-image: don't ignore symlinks to raw images
Since 5c6bb289990ba53898cbc62db6e732ecb9dc87ac image_discover() uses
chaseat() to chase the path to the image. This however breaks the raw
image check in image_make() as "path" is now not the symlink itself, but
the symlink target.
The endswith(path, ".raw") check is now performed on "/.../foo.squashfs"
instead of "/.../foo.raw", making it false and thus ignoring the image
symlink completely.
Address this by also checking if the pretty name is set - if so, and the
path is a regular file, the caller must've been image_find() or
image_discover() which already checked if the original path ends in .raw
and is a regular file.
repart,dissect: explicitly support DDIs that are both signed *and* encrypted (#43009)
This is a pretty relevant usecase: preparing an image on some trusted
host, submitting it to some other host that authenticates it and
decrypts it, and consumes it only then.
machined: Allow user ids in open_shell for machine-dbus
Previously "machinectl shell --uid=1000 <container>" resulted in a
sucessful drop into a shell in the respective target machine. After
commit a9e9288288567beae57337ae903dd3b6c774001c this is no longer the
case.
udev-util: bound leading whitespace skip in udev_replace_whitespace
The function documents that at most 'len' bytes are read from 'str',
but the leading whitespace skip used strspn(), which is bounded only
by a non-whitespace byte or a NUL. ata_id passes the space padded,
non-NUL-terminated ATA IDENTIFY fields, so an all-blank model reads
past the buffer. Use strnspn() to cap the skip to 'len'.
Like strspn(), but reads at most 'n' bytes from the input. strspn()
is bounded only by a non-matching byte or a NUL, so it over-reads a
buffer that is all matching bytes and not NUL terminated within 'n'.
For various cases it is interesting to both sign and encrypted a file
system, for example to prepare it on one host and provide it to another.
Let's explicitly support preparing this in systemd-repart via setting
both Verity= and Encrypt=.
sysext: validate work directory metadata before removal
unmerge_hierarchy() joined the persisted work_dir value directly with
--root=. An empty value therefore resolved to the root itself and was passed
to rm_rf().
Require the decoded metadata to name a non-empty, safe, normalized relative
path before constructing the removal target. Add coverage using a disposable
root with deliberately emptied metadata.
shift-uid: close consumed directory fds on early return
recurse_fd() consumes each directory fd passed to it, but it has no
cleanup set up until after take_fdopendir() succeeds.
Errors and skipped procfs, sysfs, or read-only subtrees therefore leak the
incoming fd.
path_is_root_at() returns a negative errno when it cannot determine whether
a target is the root file system. rm_rf_at() treats those errors as a
negative answer and continued with destructive operations.
Propagate root-check errors before modifying the target, while preserving
REMOVE_MISSING_OK for an absent path.
vconsole: reject empty layout during keymap conversion with error
A leading-dash console keymap or leading-comma X11 layout can produce an
empty layout while converting between the two formats. find_converted_keymap()
then asserts on it, allowing malformed input to abort callers such as localed.
Return EINVAL for an empty derived layout instead.
resolved: preserve unchanged question on asymmetric redirect
dns_question_cname_redirect() returns no replacement when a question
already matches a CNAME target or a DNAME does not apply. If only one of
the UTF-8 and IDNA questions redirects, dns_query_cname_redirect() then
installs NULL for the unchanged side.
Keep a reference to the original question on whichever side does not
redirect, and update the existing asymmetric DNAME test to verify it.
MHDDaemonWrapper_free() unconditionally unrefs both event sources, but
setup_microhttpd_server() could return before initializing either pointer.
Zero-initialize the wrapper so cleanup after MHD startup failures safely
unrefs NULL.
journal-remote: remove disabled compression entry before freeing
compression_config_put() frees the COMPRESSION_NONE value when a later
algorithm is selected, but lefts the map entry pointing at it. Remove the
entry first so subsequent parsing, mangling, and teardown cannot access or
free stale memory.
network: cancel netlink calls for detached requests
Store the asynchronous netlink slot in the Request object and release it
when the request is detached. This cancels the pending callback instead of
keeping the request alive until its reply arrives.
The multipath route comparator looks up every nexthop in the first
route and hence compares each entry with itself. Compare both ordered
nexthop sets instead, including weights, as the hash function does.
Paul Meyer [Tue, 14 Jul 2026 11:26:56 +0000 (13:26 +0200)]
tpm2-util: keep the measurement log's torn-write marker intact
The userspace measurement log carries a sticky-bit marker while a writer
is between updating a measurement register and appending the matching
record, so that a writer dying in between leaves the log detectably
incomplete.
However, the next successful writer used to clear the marker again after
appending its own record, erasing the evidence that an earlier writer
had died and the log is still missing a record. Keep the marker set in
that case instead; the new record is appended and synced regardless. The
marker likewise stays set if the measurement itself fails, so that any
non-clean completion remains flagged: a spurious flag on a failed but
harmless measurement is preferable to erasing evidence of a real gap,
and PCR replay stays authoritative either way.
Finally, warn when systemd-pcrlock loads a marked log, so the resulting
PCR validation failures come with a hint at their cause.
Before, the field listing ignored the unit filter and listed unrelated
units, while the normal journal query applied the filter.
sd_journal_query_unique() cannot represent journalctl filters such as
unit, boot, time, cursor, or grep filters. Walking the journal line by
line would make field listing linear in the number of entries and
duplicate unique-value handling.
Keep the scope-option predicate next to add_filters(), and reject field
listings combined with options that actually limit the journal.
Display-only options such as --pager-end are left alone.
The target parameter length is added to two headers and aligned before
allocation. Check each operation and reject payloads that cannot be
represented by the device-mapper ioctl header.
gaih_addrtuple.addr has room for an IPv6 address, while IPv4 uses only
the first struct in_addr. Express the remaining range directly in terms
of that type so static analysis can prove the access remains within the
array.
* 88c420c621 Install new files for upstream build
* 5b7f67b86a Update changelog for 261.1-3 release
* 3f91c8567c NEWS: update about removal of sd-tpm dep
* e13ba94737 Drop obsolete TODO
* b0c5def366 systemd: downgrade systemd-tpm and mount to recommends
* e9eb859a3a systemd-tpm: depend on tpm-udev for rules and tmpfiles.d
* 42f74bb110 Drop obsolete comment from d/control
* fc40986137 debian/extra/network: use NamePolicy=keep mac on USB wifi devices
Kai Lüke [Wed, 15 Jul 2026 07:09:13 +0000 (16:09 +0900)]
oci-util: Don't fall back to default registry for explicit registries
When a reference specifies a registry like ghcr.io or quay.io for which
we don't have a registry config file, the fallback forced the default
registry which is wrong.
Only use the default as fallback when there is no explicit registry.
config_add_entry_loader_auto() returns NULL when the Windows loader
is missing. With reboot-for-bitlocker enabled, the caller still
assigned e->call and could dereference NULL.
This can happen when reboot-for-bitlocker is enabled but the ESP does
not contain EFI/Microsoft/Boot/bootmgfw.efi.
Before:
systemd-boot could dereference NULL while building the menu
Aritra Basu [Sun, 7 Jun 2026 06:04:11 +0000 (02:04 -0400)]
network: add IPv4ProxyARPAddress= and consolidate proxy ARP/NDP handling
This adds an IPv4 counterpart to `IPv6ProxyNDPAddress=` for adding
manual entries to the kernel's IPv4 neighbour proxy table (check via
`ip -4 neighbour show proxy dev <dev>`). systemd-networkd only exposed
`IPv4ProxyARP=` for per-interface `proxy_arp` sysctl (automatic proxy
ARP) with no way to manage manual entries from a .network file.
To avoid duplicating the IPv6 proxy NDP code path, both families are
now combined into a single new `networkd-neighbor-proxy` module. The
IPv6 behaviour is preserved: `IPv6ProxyNDPAddress=` still implies
`IPv6ProxyNDP=yes` unless `IPv6ProxyNDP=` is explicitly disabled and
entries are still dropped if the kernel has no IPv6 support.
The same rule is applied to `IPv4ProxyARPAddress=`. It implies
`IPv4ProxyARP=yes` when the sysctl is not explicitly set and has no
effect if `IPv4ProxyARP=` has been set to false.
This keeps the user model symmetric and predictable across both
families: a single per-address setting that turns on the matching
per-interface sysctl automatically, while still letting system
administrators opt out by setting the boolean explicitly to false.
Note that the IPv4 manual NTF_PROXY entries installed here would
actually function without `proxy_arp` (unlike IPv6, where `proxy_ndp`
gates the manual entries); the implication is kept for symmetry with
`IPv6ProxyNDPAddress=` and is now called out explicitly in the man
page, together with the fact that enabling `proxy_arp` also activates
interface-wide automatic proxy ARP for routed-toward addresses on
connected subnets.
Parser-time validation rejects addresses the kernel would refuse:
the ANY/null address for both families, IPv4 and IPv6 multicast
and the IPv4 limited broadcast 255.255.255.255.
shared/copy: simplify argument convention for reflink_range
reflink_range() would treat size==UINT64_MAX as "copy everything", but
only if the offsets were 0. There are only two callers: one always passes
a fixed size, the other translated size==UINT64_MAX to 0. Make things
more uniform by always treating size==UINT64_MAX the same as size==0.
Paul Meyer [Mon, 13 Jul 2026 07:47:40 +0000 (09:47 +0200)]
tpm2-util: initialize NvPCRs on first extension
Both callers of tpm2_nvpcr_extend_bytes() duplicate the same dance: on
-ENETDOWN, i.e. when the NvPCR isn't anchored yet because
systemd-tpm2-setup hasn't run, they acquire the anchor secret,
initialize the NvPCR, and extend again. Move this into
tpm2_nvpcr_extend_bytes() itself. The only caller-specific bit, whether
the anchor secret shall be synced to /var, is passed in as a parameter.
Reflinking is a safe optimization whenever the source and destination are
regular, seekable files on a filesystem that supports cloning. Requiring each
caller to opt in through COPY_REFLINK adds flag plumbing without changing the
necessary fallback behavior.
Drop COPY_REFLINK, renumber the remaining flags, and have copy_bytes_full()
unconditionally try FICLONE or FICLONERANGE before entering its normal copy
loop. Isolate the clone attempt and its file-offset bookkeeping in a helper so
copy_bytes_full() only has to handle its cloned, unavailable, and error results.
A successful clone therefore completes in one operation, while an unsupported
or rejected clone falls back to progress-limited copy_file_range(), sendfile(),
or buffered copying. Keep the public reflink helpers for operations that
specifically require clone semantics.
Existing test-copy coverage exercises both bounded and unbounded regular-file
copies through copy_bytes_full().
systemd-repart currently forces newly created image files into NOCOW mode.
That prevents files from being reflinked into the image, making image builds
slower and increasing their disk usage on filesystems that support cloning.
Add a tristate --cow= option. By default, leave the filesystem or parent
directory COW policy unchanged. With --cow=yes, explicitly enable COW; with
--cow=no, retain the previous behavior of forcing NOCOW. Add XO_COW as the
counterpart to XO_NOCOW so xopenat_full() applies either policy while retaining
its normal creation-error cleanup.
Document the new option and extend TEST-58-REPART to verify inherited COW and
NOCOW policies as well as explicit COW and NOCOW overrides. Compare the unset
behavior with the filesystem default so the test also works on nodatacow
mounts, and skip it when the inode attribute is unsupported.
Commit 04f19d673587 ("udev: Add /dev/media/by-path symlinks for media
controllers") added the rule file and the feature was announced in the
v256 NEWS, but the file was never added to the rules.d/meson.build
install list, so no build has ever shipped it and the advertised
/dev/media/by-path/ symlinks are never created.
sd-dhcp-relay: fix off-by-one when discarding BOOTREQUEST messages by hops count
According to RFC specifications
```
RFC 1542 section 4.1.1 states:
The relay agent MUST silently discard BOOTREQUEST messages whose 'hops'
field exceeds the value 16."
```
"Exceeds the value 16" means hops > 16, i.e. a message that arrives with
hops == 16 is still valid and must be relayed (after which its hops field
becomes 17). The code used ">= 16", which silently dropped a valid message
that had legitimately traversed exactly 16 relay agents, one hop too early.
This matches the wording of the adjacent comment, which already says
"exceeds the value 16".
resolved: fix spurious BrowseServices add/remove flapping with ifindex=0 (#42982)
## Problem
A `BrowseServices` subscription with `"ifindex":0` (browse all
interfaces) receives a continuous flap of `added`/`removed` events for a
service that is still present — within a second, and with no goodbye
packet involved.
## Root cause
For `ifindex==0`, `mdns_browser_revisit_cache()` looked up each mDNS
scope's cache separately and called `mdns_manage_services_answer()`
**once per scope**. That function derives `removed` events by diffing
the browser's *global* discovered-service list (all interfaces, filtered
only by owner family) against the single answer it's handed. So with ≥2
mDNS-relevant interfaces of the same family, a service present on
interface A isn't in interface B's answer and is spuriously removed
while B is reconciled, then re-added on the next revisit tick.
## Fix
Accumulate the pruned cache answers from every matching mDNS scope into
one combined `DnsAnswer` and reconcile **once**, so removals diff the
global list against the union across interfaces. Items are merged with
`dns_answer_add_full()` (not `dns_answer_extend()`, which defaults each
item's `until` to `USEC_INFINITY` and would skew the RFC 6762 §5.2
TTL-maintenance schedule). The single-interface (`ifindex>0`) path is
unchanged.
## Test
`TEST-89-RESOLVED-MDNS.sh` gains `testcase_browse_ifindex_zero_no_flap`:
it adds a service-less dummy mDNS link to guarantee ≥2 same-family
scopes (the flap precondition), browses `ifindex=0`, waits for
discovery, then asserts **zero** `removed` events while every publisher
stays up. The subscription uses `varlinkctl --timeout=infinity`, since
it sits idle after discovery and the default 45s idle timeout would
sever it (and the assertion) mid-observation.
## Testing status
Builds clean; `shellcheck -x` clean. First CI round: `TEST-89`
(including this testcase) passed on all mkosi platforms; the failing
jobs all traced to unrelated flakes/infra.
U-Boot currently does not implement EFI_MEMORY_ATTRIBUTE_PROTOCOL
even when reporting EFI version >= 2.10. Consequently, systemd-boot
emits a warning on every boot when running on U-Boot firmware.
The absence of EFI_MEMORY_ATTRIBUTE_PROTOCOL is a current
U-Boot limitation and not a condition users can remedy.
Furthermore, the EFI specification does not require all
firmware advertising EFI 2.10 or newer to implement the
protocol. As a result, the warning provides little value on
U-Boot systems while causing log_wait() to impose a 2.5-second
boot delay.
Downgrade the message to LOG_DEBUG, this keeps the diagnostic
available for debugging purposes without penalizing normal
boot time.
sysupdate: Add ListFeatures() and ListTargets() varlink methods (#42900)
Following on from adding the basic varlink scaffolding to sysupdate,
let’s varlinkify a couple of the D-Bus methods. Because varlink doesn’t
have a concept of object paths, the D-Bus path structure which allows a
target to be selected has been squashed down to a target argument for
each relevant method.
Varlinkify the way to list targets, and also the way to list features
because that was simple to do at the same time.
More methods need varlinkifying in the future, but let’s do it in small
and manageable chunks.
string-util: replace version_is_valid()/version_is_valid_version_spec() by a common call
Let's take inspiration from string_is_safe() and take a flags field that
allows fine tuning the validation.
Then port over all current users of either function to the new logic.
Note that this *does* change behaviour in various cases:
1. Generally: we'll now always accept the full UAPI.10 alphabet,
including the "~" and "^" characters. As far as I can see there's no
downside to this liberalization as none of the current consumers of
the two functions uses these characters for anything else.
2. systemd-analyze compare-version will now accept version strings with
"_" and "+" without complaining. I see no downside here, it just
normalizes these debugging tools, to make them accept what most our
other tools accept.
3. "bootctl link" will not accept empty version strings anymore
Which is a bugfix I guess.
4. vpick will now refuse "_" and "+" in version strings. It kinda
already did, because when parsing versions from filenames it uses "_"
and "+" as name, architecture and attempt counter separators. We now
systematically refuse it everywhere else in vpick too. This is hence
a clean-up.
string-util: reorder characters in version charset
Let's bring the version string character set into a systematic order,
matching the order in which they appear and are defined in the UAPI.10
specification text.
This makes it easier to compare the relevant functions.
Allowing this apparently has been cargo-culted from my initial sysupdate
PR, but Claude and me could not find a single other software package
that uses "," as a character within version strings. Hence, let's remove
this, even though this is a compat breakage of a kind, in the hope
nobody notices. We can easily restore this if this later shows to be an
issue for people.
The RTCTimeUSec property was being displayed in local time format
when using 'timedatectl show', while 'timedatectl status' correctly
displayed it in UTC format. This inconsistency was due to the
bus_print_all_properties() function using TIMESTAMP_PRETTY style
for all timestamps, which formats in local time.
This fix adds special handling for RTCTimeUSec to use TIMESTAMP_UTC
style, ensuring consistent UTC display across both commands.
dissect-image: don't assert() on partition geometry from blkid
The per-partition loop in dissect_image() reads the start and size (in
512-byte sectors) of each partition from libblkid and guards the
following byte conversions with assert(). Images sizes are input,
rather than programming, so return an error instead of asserting.
homework-luks: add new key slots before destroying old ones
home_passwd_luks() rotates the LUKS key slots with a single loop that
destroys slot i before adding its replacement at the same index. If
adding the replacement fails (e.g.: argon2 OOMs), slot i is left
destroyed with no replacement and no rollback. If the user has only
one password, its only key slot is now gone and the home directory can
no longer be unlocked.
Rotate in two passes instead: first add every new password into a free
slot (CRYPT_ANY_SLOT), and only once all adds succeed, destroy the old
slots. If an add fails, roll back the slots added so far and return
with the pre-existing slots untouched, so at least one valid key slot
always remains for each password the user holds.
Philip Withnall [Mon, 13 Jul 2026 14:55:41 +0000 (15:55 +0100)]
sysupdate: Allow multiple documentation URLs for a feature
Change the varlink API for Feature structs to allow multiple
documentation URLs for them, to match what systemd already does for
units etc.
This is a deviation from what the sysupdated D-Bus API allows and, for
the moment, from what’s supported internally by sysupdate. Internally it
continues to support 0-1 URLs for now.
But by defining the API as a strv, multiple URLs can be supported in
future without API breaks.
Philip Withnall [Thu, 9 Jul 2026 12:55:01 +0000 (13:55 +0100)]
sysupdate: Run ListFeatures in offline mode
Historically, the ListFeatures API (in both D-Bus and now varlink) was
run without an `--offline` flag. This appears like it’s an oversight, but
actually `verb_features()` always unconditionally loaded in offline
mode.
In any case, there doesn’t appear to be a reason for the context to be
online (i.e. for it to check sources for available updates). Features are
defined in local config files and are loaded by `read_features()`, which
is called in both offline and online mode.
When the varlink ListFeatures API was added, it was put into online mode
in order to match the lack of `--offline` argument in the existing D-Bus
API implementation. Change both of them to be explicitly in offline mode.
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Philip Withnall [Mon, 22 Jun 2026 19:20:20 +0000 (20:20 +0100)]
sysupdate: Downgrade an info to a debug log message
Since we now enumerate all targets on a varlink call (to validate the
requested target), this message gets printed multiple times in the log.
It’s not really necessary, so downgrade it to a debug message.
Previously, files listed in 'sources' were built twice:
once when building the main binary, and again when building the
statically linked one.
This change ensures that all object files from the main binary are
reused when building the static binary. Hence, the only step now
necessary for the static binary is linking the object files.
Jonas Dreßler [Mon, 6 Jul 2026 16:23:51 +0000 (18:23 +0200)]
repart: Properly pre-calculate auto size of images
When passing --size=auto to repart, it will pre-calculate the image size and
resize the image to that size before partitioning. Currently, that fails when
passing a large grain size, complaining that the auto-sized image is too small
to fit the data.
The reason for this is that the current code simply assumes the GPT metadata
size taken away from the usable size by fdisk is static (1044KiB), when it
actually is more complicated than that:
There's two ranges of GPT metadata: One at the beginning of the image, and one
at the end of the image. And there's the first usable block that is defined by
fdisk when creating the partition table.
The static value of 1044KiB usually works, because fdisk sets the first usable
block to 1MiB (so 1024KiB), leaving 20KiB of leeway for the secondary GPT at
the end of the image.
Now as soon as the first partition starts at an offset higher than 1024KiB, we
lose the 20KiB leeway for the secondary GPT, and the partitions will no longer
fit.
What we should do, is first of all round up to the grain size instead of 4096
(as that's the minimum offset our first partition will start at), and second of
all properly subtract the secondary GPT at the end.
Also confirm we don't regress on this anymore by adding a test that uses a 2MiB
grain size, breaking the old code.
projects / thirdparty / systemd.git / log
