The branch with configure_file() was broken: meson doesn't know that
this file is a prerequisite for other targets, so partial rebuilds were broken.
Easy reproducer:
git mv .git{,.no}
touch meson build && ninja -C build src/basic/libbasic.a
rm build/version.h
ninja -C build src/basic/libbasic.a
Using vcs_tag() also in that case makes meson always build the file.
(Combined with the issue fixed in previous commit, I was encountering
failed builds quite often.)
With git-worktree, .git is just a file that specifies where
the parent git directory is. All the git information is available
in a git worktree, so it should be treated the same as a checkout
with a .git directory.
Lars Ellenberg [Wed, 7 Feb 2024 12:12:50 +0000 (13:12 +0100)]
fs-utils: new wrapper fd_reopen_propagate_append_and_position()
We may want to propagate O_APPEND, or (try to) keep the current file position,
even if we use fd_reopen() to re-initialize (and "unshare") other file
description status.
For now, used only with --pty to keep/propagate O_APPEND (and/or) position
if set on stdin/stdout.
If we re-open stdout and "drop" the O_APPEND,
we get rather "unexpected" behavior,
for example with repeated "systemd-run --pty >> some-log".
If someone carefully pre-positioned the passed in original file descriptors,
we avoid surprises if we do not reset file postition to zero.
fcntl F_GETFL first, and propagate O_APPEND if present in the existing flags.
Frantisek Sumsal [Tue, 12 Mar 2024 12:11:16 +0000 (13:11 +0100)]
test: split logs from each test into separate files if requested
If both $ARTIFACT_DIRECTORY and $SPLIT_TEST_LOGS are set, split the
output from each test into a separate log file, so we don't have to load
one ginormous log file when checking the results.
Frantisek Sumsal [Tue, 12 Mar 2024 13:49:55 +0000 (14:49 +0100)]
ci: reduce ASLR entropy
The latest GH Action runners started using 32-bit entropy for ASLR,
which makes it incompatible with llvm-14. This was fixed in later llvm
releases, but these aren't available on Ubuntu Jammy (22.04). Let's
reduce the ASLR entropy to 28-bit, which should make llvm happy again,
until the issue is resolved.
Luca Boccassi [Wed, 6 Mar 2024 01:12:19 +0000 (01:12 +0000)]
ukify: really add default .sbat for UKIs
This was lost on refactor, and only addons had a default uki
line in the .sbat. Add it back, and differentiate between the
default for UKIs vs the default for addons, so that they can
be revoked separately. These are only defaults and users are
encouraged to provide their own.
stat-util: add explicit helpers for checking if stat/statx is initialized
This is a follow-up for 945a8210c770801c8492eda03b6e9af3ec5d03a3 and
makes the st_dev check generic, so that we can reuse it some other
places. It also incorporates the non-NULL check now, to be a
comprehensive one-stop solution.
The helper is static inline so that compilers can optimize the redundant
checks away in case it is combined with other checks.
Ross Burton [Mon, 11 Mar 2024 16:41:02 +0000 (16:41 +0000)]
tests/run-unit-tests: add option to skip tests
In automated QA environments there may be tests that are known to fail,
and being able to skip them is useful to remove known failures from the
test log.
Adrian Vovk [Thu, 7 Mar 2024 18:38:26 +0000 (13:38 -0500)]
homework-cifs: Pass password via fd
Pass the password into mount.cifs via a file descriptor, rather
than putting it into a plain-text file in /tmp. This uses the $PASSWD_FD
environment variable, which is undocumented but has existed since
forever (initial commit from 2010 [1] has it already)
Yu Watanabe [Fri, 1 Mar 2024 03:10:49 +0000 (12:10 +0900)]
network/dhcp-server: save leases in state directory
Then, we can read the lease file on restart, and the DHCP server will be
able to manage previously assigned addresses.
To save leases in the state directory /var/lib/systemd/network/, this
adds systemd-networkd-dhcp-server.service, and by default
systemd-networkd does not start the DHCP server without the heler
service started.
Yu Watanabe [Mon, 11 Mar 2024 16:47:17 +0000 (01:47 +0900)]
sd-dhcp-server: refuse invalid hostname in request
Currently, the received hostname is not used for assigning an address to
the host, or options in the subsequent reply message. But, the parsed
hostname is exposed through DBus, and possibly Varlink in the future.
Let's ignore spurious hostname.
Yu Watanabe [Mon, 11 Mar 2024 16:32:03 +0000 (01:32 +0900)]
dhcp-option: refuse control and non-UTF8 characters in string option
We oftem save parsed DHCP options into a file, or expose them
through DBus or Varlink. In such case, control characters or non-UTF8
characters may cause many kind of unexpected errors. In general, a DHCP
message that have string options with spurious characters is mostly
malformed or broken. Let's refuse them.
This also makes dhcp_option_parse_string() do not free 'ret' argument,
to follow our usual coding style. So, callers now need to free the
pre-exisitng string if necessary.
Then, this introduces systemd-networkd-persistent-storage.service.
systemd-networkd.service is an early starting service. So, at the time
it is started, the persistent storage for the service may not be ready,
and we cannot use StateDirectory=systemd/network in
systemd-networkd.service.
The newly added systemd-networkd-persistent-storage.service creates the
state directory for networkd, and notify systemd-networkd that the
directory is usable.
Mike Yuan [Mon, 11 Mar 2024 10:27:50 +0000 (18:27 +0800)]
stat-util: struct stat could be initialized with (st_mode & S_IFMT == 0)
For anonymous inodes, the result would be 0, but
the struct stat is initialized obviously.
So let's switch to st_dev for the check, which
is guaranteed to be non-zero.
Also this is completely unnecessary for statx(),
since we check stx_mask first and that on its own
denotes that the struct is initialized.
polkit: turn "interactive" flag to polkit APIs into a proper flags field (#31715)
This adds for both the D-Bus and the Varlink flavours of our polkit
client api a flags parameter. And then folds the "bool interactive" flag
that the D-Bus version so far had, into a flag, and also adds support
for it in the Varlink API.
Since this means the Varlink API gained another parameter, let's do what
we already did for the D-Bus API and add a _full() version of the API
that has the flags and the good_uid parameter, and one without both.
This fixes the following error building against a bleeding edge kernel.
```
../src/basic/meson.build:234:8: ERROR: Problem encountered: Unknown filesystems defined in kernel headers:
Filesystem found in kernel header but not in filesystems-gperf.gperf: PID_FS_MAGIC
```
Mike Yuan [Mon, 11 Mar 2024 07:29:41 +0000 (15:29 +0800)]
gpt-auto: ignore fstab_has_node failure
After 8a1326581d9b066377f8d9f2d58e1bdfd8b645d0,
we always check whether there're mounts under
/boot/ or /efi/ first. Let's relax the check
for fstab_has_node hence, since on initrd-less
systems it might produce wrong results.
Luca Boccassi [Sat, 9 Mar 2024 12:42:32 +0000 (12:42 +0000)]
semaphore: move back to autopkgtest master branch
Instead of fixing the commit, we can workaround the adduser issue by
simply creating a user manually beforehand, which means the broken
codepath in autopkgtest is not taken. We can remove it once it's
fixed upstream, which is in progress:
Ronan Pigott [Fri, 8 Mar 2024 21:48:03 +0000 (14:48 -0700)]
resolved: also reply NOTIMP when refusing a query based on RR type
In some cases we refuse a query based on the RR type, mostly old
deprecated types. Let's return NOTIMP in this case, which best
communicates why the query failed.
Chandra Pratap [Fri, 8 Mar 2024 14:37:29 +0000 (20:07 +0530)]
extract-word: increase test-extract-word coverage for unicode inputs
In the current testing scheme in test-extract-word, we only
have two test cases covering unicode strings. Improve upon
this by adding more cases for the same.
Daan De Meyer [Fri, 8 Mar 2024 18:12:16 +0000 (19:12 +0100)]
mkosi: Add BuildSourcesEphemeral=yes
Required to make sure that any changes packaging specs make to the
source files are thrown away after the build so they don't mess with
the source tree.
Ronan Pigott [Wed, 6 Mar 2024 02:03:16 +0000 (19:03 -0700)]
resolved: don't cache NXDOMAIN for SUDN resolver.arpa
The name resolver.arpa is reserved for RFC9462 "Discovery of Designated
Resolvers" (DDR). This relies on regular dns queries for SVCB records at
the special use domain name _dns.resolver.arpa. Unfortunately, older
nameservers (or broken ones) won't know about this SUDN and will likely
return NXDOMAIN. If this is cached, the cache entry will become an
impediment for any clients trying to discover designated resolvers
through the stub-resolver, or potentially even sd-resolved itself, were
it to implement DDR.
The RFC recommendation is that "clients MUST NOT perform A or AAAA
queries for resolver.arpa", and "resolvers SHOULD respond to queries of
any type other than SVCB for _dns.resolver.arpa. with NODATA and queries
of any type for any domain name under resolver.arpa with NODATA." which
should help avoid potential compatibility issues. This enforces that
condition within sd-resolved, and avoids caching any such erroneous
NXDOMAIN.
The RFC also recommends requests for this domain should never be
forwarded, to prevent authentication failures. Since there isn't much
point in establishing secure communication to the local stub, we still
allow SVCB to be forwarded from the stub, in case the client cares to
implement some other authentication method and understands the
consequences of skipping the local stub. Normal clients are not
expected to implement DDR, but this change will protect sd-resolved's
own caches in case they try.
Although A and AAAA are prohibited, I think validating resolvers
might reasonably query for dnssec records, even though the resolver.arpa
zone does not exist (it is declared to be a locally served zone). For
this reason, I have also added resolver.arpa to the builtin dnssec NTA.
Ronan Pigott [Thu, 7 Mar 2024 21:27:52 +0000 (14:27 -0700)]
resolved: refuse queries with no suitable scope
In some cases there is no configured server to answer a given question,
because all scopes refused the query. In this case we currently return
rcode SERVFAIL.
In dns it is customary for authoritative nameservers to return REFUSED
where the question is outside of their authority. This is better than
SERVFAIL because it informs the client that they aren't likely to get an
answer out of us anytime soon, and either the configuration, or the
query, need to change.
Similar logic invites us to use use the rcode REFUSED on the stub if we
aren't configured with any suitable scope for this question.
kernel-install: support full set of config files and drop-ins
This brings the handling of config for kernel-install in line with most of
systemd, i.e. we search the set of paths for the main config file, and the full
set of drop-in paths for drop-ins.
various: use new config loader instead of config_parse_config_file()
This means the main config file is loaded also from /run and /usr.
We should load the main config file from all the places where we load drop-ins.
I realize I had a giant blind spot: I always assumed that we load config files
from /etc, /run, /usr/local/lib, /usr/lib. But it turns out that we only used
those paths for drop-ins. For the main config file, we only looked in /etc. The
docs actually partially described this behaviour, i.e. most SYNOPSIS sections
and some parts of the text, but not others.
This is strange, because 6495361c7d5e8bf640841d1292ef6cfe1ea244cf was completely
bogus with the behaviour before this patch. We had a huge discussion before it
was merged, and clearly nobody noticed this. Similarly, in the previous version
of the current pull request, we had a long discussion about the appropriate
order of directories, and apparently nobody noticed that there was no order,
because only looked in one directory. So the blind spot seems to have been
shared.
Also, systemd-analyze cat-config behaved incorrectly, i.e. its behaviour matches
the new behaviour.
Possibly, in the future it'll make it easier to add support for --root.
projects / thirdparty / systemd.git / log
