Commit f8fa4222c9 added extra logic to
account for the size of the initrds before allocating the required memory pages,
but missed checking if the initrds were empty before allocating memory.
Extend the initrd options only when the initrd is valid, and add extra
check for validating that there is at least one valid initrd file before
allocating memory, fixing the following crash when booting with empty
initrd files:
src/boot/log.c:30@efi_assert: systemd-boot: Assertion
'BS->AllocatePages(type, memory_type, n_pages, &addr) == EFI_SUCCESS'
failed at /usr/src/debug/systemd-boot/257.8/src/boot/util.h:95@xmalloc_pages, halting.
Fixes: f8fa4222c9
("boot: Make initrd_prepare() semantically equivalent to combine_initrds()")
(cherry picked from commit a40f4ba1a8)
When systemd-resolved is used in resolv.conf foreign mode, reloading the
systemd-resolved daemon does not read resolv.conf when not touched since
the last time resolved read the file. Since the DNS servers list is
flushed during a reload, resolved forgets about the servers
in /etc/resolv.conf
When reloading the daemon it is expected that all configuration is
reloaded regardless of the disk timestamps.
(cherry picked from commit ca1d2ae5d6)
```
==19541== 8 bytes in 1 blocks are still reachable in loss record 1 of 3
==19541== at 0x4841744: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==19541== by 0x50125C9: strdup (strdup.c:42)
==19541== by 0x4C55925: getusername_malloc (user-util.c:154)
==19541== by 0x1121D6: parse_argv_sudo_mode (run.c:1098)
==19541== by 0x123B13: run (run.c:3032)
==19541== by 0x124198: main (run.c:3100)
==19541==
==19541== 11 bytes in 1 blocks are still reachable in loss record 2 of 3
==19541== at 0x4841744: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==19541== by 0x50125C9: strdup (strdup.c:42)
==19541== by 0x112A9C: parse_argv_sudo_mode (run.c:1182)
==19541== by 0x123B13: run (run.c:3032)
==19541== by 0x124198: main (run.c:3100)
```
(cherry picked from commit 7599d26436)
The offset must be specified first, 'whence' second. Fix that.
Except for one case this fix doesn't actually fix any real bug, since
SEEK_SET is defined as 0 anyway, hence the swapped arguments have no
effect.
The one exception is the MTD smartmedia code, which I guess indicates
that noone has been using that hw anymore in a long time?
(cherry picked from commit f5452477d4)
We try to read again from the beginning, hence let's seek back.
Apparently efivarfs doesn't strictly require this, but it's really weird
that it doesn't.
(cherry picked from commit dbc25d84ae)
writev() returns the full size, not just the payload size, hence always
add sizeof(attr) where necessary.
Let's also change a couple of "4" into sizeof(attr) all over the place,
to make clear what they are about.
Fixes: #39695
Follow-up for: 9db9d6806e
(cherry picked from commit ab69a04600)
We shouldn't ask glibc to keep the old data around (which realloc() is
about), given we overwrite it entirely anyway. Let's hence speed things
up here, and allow glibc to just allocate a new block for us (and
shorten the code a bit)
(cherry picked from commit 88e26303ce)
Make bus acquisition conditional in verb_edit() and verb_cat(), following
the same pattern used in verb_enable(). When install_client_side() returns
non-zero (indicating --global, --root, offline, or similar scenarios), skip
acquiring a D-Bus connection and perform all operations client-side.
Changes:
- Only acquire bus when install_client_side() returns NO
- Use mangle_names() instead of expand_unit_names() in client-side mode
- Pass force_client_side flag based on bus availability
- Skip bus-dependent operations (need_daemon_reload, etc.) when bus is NULL
This allows 'systemctl edit --global' and 'systemctl cat --global' to work
correctly, fixing the regression introduced by commit d77d42ed3a.
Test cases added to verify:
- Creating and editing global user units with --runtime
- Reading global units with cat --global
- Proper detection and rejection of masked units in client-side mode
- Tests use /run/ instead of /etc/ for safer temporary testing
Fixes https://github.com/systemd/systemd/issues/31272
(cherry picked from commit ebd222b1c8)
When operating in client-side mode (force_client_side=true), unit_find_paths()
now checks if the unit file is masked (symlinked to /dev/null or empty) and
returns -ERFKILL, matching the behavior of the server-side path.
This centralizes masked unit detection in one place, making it consistent
across both client-side and server-side operations.
(cherry picked from commit 423a8ffccc)
When VirtIO VSOCK device is not present, IOCTL_VM_SOCKETS_GET_LOCAL_CID
returns VMADDR_CID_LOCAL/1, and we issue a hint to connect to vsock%1.
This does not work. Filter out VMADDR_CID_LOCAL and VMADDR_CID_HOST,
those are not real addresses that can be used from the outside.
(cherry picked from commit 492ae9ec4e)
Commit 0bac1ed242 ("tree-wide: Fix constness issues with newer glibc")
split a temporary var in get_usb_specifier to const and non-const
versions, but missed converting a couple of uses. This means we get
names of with components of:
port: uN
config: cC.I
interface: iC.I
instead of:
port: uN
config: cC
interface: iI
This results in net names like enu1c1.0i1.0, as we also no longer hit
the config==1 and interf==0 elision cases.
Change the config portion handling to start from the correct position,
and the earlier check for NULL.
(cherry picked from commit b10b4303c6)
The completion fails to complete to paths for verbs that need them when
a --param is the previous word, e.g.:
portablectl attach --runtime <tab>
fails to complete to paths
(cherry picked from commit 2abb433ca1)
The path--as a whole--is no longer being opened with O_DIRECTORY, hence
the assertion is not valid.
Reproducer:
# SYSTEMD_RELAX_ESP_CHECKS=1 bootctl list --esp-path=`pwd`/README
Assertion 'S_ISDIR(sxa.stx_mode)' failed at src/shared/find-esp.c:295, function verify_fsroot_dir(). Aborting.
Follow-up-for: 63105f33ed
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=2417917
(cherry picked from commit 50482aced5)
Access to these methods is also managed via polkit,
but they were never added to the access list in the
D-Bus policy.
Follow-up for 907952bbc9
(cherry picked from commit 28c68dbea0)
The system may be using a stat(1) implementation that follows
BSD-style syntax. Try that as a fallback, but do fail for real
if that also fails.
(cherry picked from commit 57f911973a)
The coreutils release that included this was in 2012. Since then,
systemd requires a newer meson, mostly uses install_symlink, and
so on. Additionally, where it does use this feature, it uses the
'-r' short flag. There are implementations of 'ln' that do have
support for '-r' but no longopts. Things will still work on those
but this check will fail.
(cherry picked from commit f40bd7af20)
* 5a476a92de Use Path.relative_to instead of Path.parts
* 03f41faf9f Update doc for mkosi ssh to mention sd v256 sshd-vsock
* e054976d18 Rewrite note on fw deps of kernel modules in manpage
* a046243a50 Add python3-pefile to fedora tools conf
* 0bab54087e Fix truthy condition
* 01c38544b2 code review
* 563dca41da code review
* 6074d14c0e code review
* 7e842db3b1 Don't discard ordering of include/exclude entries in KernelModules=
* 0047b64785 build(deps): bump github/codeql-action from 4.31.2 to 4.31.6
* f13325bc4b build(deps): bump actions/checkout from 5.0.0 to 6.0.0
* 9ac31d1795 Add zram and nfnetlink to default initrd
* 4bdb47b6da verity: copy signing certificate to /usr/lib/verity.d/
* 8865afde9e Improve manpage output using pandoc lua filter (#4005)
* a425313c58 Pass environment to systemd-sysusers invocation
* 23f09d5de7 Make Ruff check for dangling whitespace
* abeed73d13 qemu: Add missing sandbox
* a05872e523 Ssh=yes maps to Ssh=always, update the manpage
* 79489534ca Warn of mkosi's sshd lacking support in distros' SElinux policy
* 696f99da43 Manpage: Update supported output formats
* 21675c4822 Silence swtpm SHA1 signature deprecation warning
* df53510982 DOC: update Ssh= to note sd-v256+ already generates sshd-vsock
* 50ccef7926 Configure default tools tree with ToolsTree=yes
* 0b701c690d opensuse: do not install openSUSE-release for Leap >= 16.0
* 4bababf421 opensuse: fix mirror URL for Leap >= 16.0
* c43123398c config.py: add config_default_proxy_exclude()
* 07a0a2be67 installer: Only mount configured state subdirs into sandbox
* d5efbf6f8c build(deps): bump github/codeql-action from 3.30.5 to 4.31.2
* 2912ed20d5 PTY forward fixes
* 66dab75259 sandbox: Drop all capabilities that don't make sense in userns
* 8fe5df4400 Respect SYSTEMD_TINT_BACKGROUND and SYSTEMD_ADJUST_TERMINAL_TITLE
* b9d208789d log: Drop konsole hack
* 7fa6eac65c Clarify that mkosi.images is not picked up in mkosi.conf.d
* 08a0f768dd vmspawn: Support BindUser=
* c7c3eaee4e mkosi-vm: Enable universe repository for ubuntu in mkosi-vm
* 688e77f1a8 vmspawn: Use --grow-image
* c47b7fe57b arch: Switch to the newly established fastly mirror
* 7db8f2cda3 Match compressed pacman packages too
* e3b3b93c50 Prevent using Overlay=yes with Format=portable
* 83a25324b3 Make sure apt sources are installed when BaseTrees= is in the mix
* f6fa2fd6a0 test: add unit tests for parents_below() utility function
* 1eafbd8d03 Don't print newlines in stash_terminal_title()
* bbaf92b50e Don't print newline after ansi sequences
* d1b84f5fb3 Look for .sdmagic before we consider a PE binary a UKI/addon
* b4efdbf14e Treat /opt the same as /usr for the sandbox
* 71ffced0f1 Rework PACKAGE_GLOBS to be a PackageManager classmethod
* 29605c2757 Various SPDX and whitespace fixes
(cherry picked from commit feb9e19ae4)
dnsmasq 2.92 started converting binary content to ascii so the previous
invalid URL is no longer invalid as it's escaped.
So the test started failing: https://bugs.debian.org/1122015
Use a '|' character instead, which is not a valid URL character but it's
also not escaped by dnsmasq.
Follow-up for 1219391c9f
(cherry picked from commit e2fbcee030)
It is a firewall service used in postmarketOS.
When it is active, then e.g. DHCP packets are filtered, and many
tests will be failed.
(cherry picked from commit 193626bd8b)
We carefully prepare a copy of a local buffer to save in device cache
and then save the buffer there instead... This leads to abort in free()
on exit (also, copied is leaked).
Reproducer:
# udevadm test /sys/block/sr0
Follow-up-for: a9559ebcbc
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=2406118
(cherry picked from commit d23a7eb4c0)
Clarify in the docs that `if-safe` is the default by noting
that in the text for it, but also moving it to the first mentioned
option.
Make explicit in `man systemd-boot` that the `secure-boot-enroll`
option is specified in the `loader.conf`
Update an outdated comment in boot.c around the same.
Signed-off-by: Colin Walters <walters@verbum.org>
(cherry picked from commit 3c85d99c79)
The gpt uuid is not the only thing that needs to match, the pattern
does as well, so mention it in the error log if present
(cherry picked from commit eb98ddd8b8)
… and also when it fails, in the few cases where this wasn't done already.
When I was debugging https://github.com/systemd/systemd/issues/39048, it
was quite confusing that we print copious information about D-Bus messages
and all other steps we're undertaking, but nothing about the sd-notify
messages which are much more important. Add some debug messages for the
cases where we didn't print anything. (The case where $NOTIFY_SOCKET is
not set in left unchanged. The variable is often left unset and we don't
need to spam logs in that trivial case.)
(cherry picked from commit 21eb636aaa)
--lines=1 means that we show the last message about the unit, whatever it is:
$ SYSTEMD_LOG_LEVEL=debug systemctl --verbose start demo2
...
Invoking 'journalctl -q --follow --no-pager --lines=1 --synchronize-on-exit=yes --unit=demo2.service' as child.
...
Directory /var/log/journal/7d16833bfa924410851e2a193bcfd4ba added.
Journal effective settings seal=no keyed_hash=yes compress=ZSTD compress_threshold_bytes=8B
...
Reiterating files to get inotify watches established.
Considering root directory '/run/log/journal'.
Considering root directory '/var/log/journal'.
Considering directory '/var/log/journal/7d16833bfa924410851e2a193bcfd4ba'.
Nov 26 17:24:35 rawhide systemd[1]: Finished demo2.service.
...
Executing dbus call org.freedesktop.systemd1.Manager StartUnit(demo2.service, replace)
...
Job for demo2.service finished.
Got result done/Success for job demo2.service.
journal: Sending message: {"method":"io.systemd.Journal.Synchronize","parameters":{"offline":false}}
Nov 27 13:05:30 rawhide systemd[1]: Starting demo2.service...
Nov 27 13:05:30 rawhide systemd[1]: demo2.service: Deactivated successfully.
Nov 27 13:05:30 rawhide systemd[1]: Finished demo2.service.
...
We obviously should only show _new_ messages, hence change to --lines=0.
This works properly after the fix in the previous commit.
Fixes#39048.
(cherry picked from commit f943ab6efb)
If --lines=0 is given, we'd skip the setup and not invoke sd_notify,
potentially blocking the caller. Change the condition for the callback
to also include that case. Since then the callback would always be
set, the 'if' statement is not necessary anymore.
(cherry picked from commit ef90afd59e)
gcry_mpi_t is defined as "typedef struct gcry_mpi *gcry_mpi_t;".
When const is applied to this type, it resolves to
"struct gcry_mpi *const" instead of what we expect ("const struct gcry_mpi *").
So we end up with a const pointer to a mutable object instead of a mutable
pointer to a const object. Since the pointer passed to the function
is copied regardless, making it const has zero benefit.
You'd think we could instead stop using gcry_mpi_t and replace it with
"const struct gcry_mpi *", except that gcrypt leaked this mess into its
api, so it expects const pointers to mutable objects as well, which means
we can't take pointers to const objects as arguments, as we'd discard the qualifier
when calling a gcrypt function.
To avoid confusion, let's drop the const qualifiers from the gcry_mpi_t arguments.
(cherry picked from commit c421392811)
- TIME_T_MAX is always time_t so the check in time-util.c was redundant
- quota-util.c was zero-ing a pointer
- smbios11.c was calling sizeof(pointer)
(cherry picked from commit 33b0642e6a)
In the CentOS Hyperscale SIG, we maintain a backport of systemd with
its own rpm spec forked from rawhide.
Let's make it easy to build upstream rpms using the Hyperscale spec
by adding a mkosi hyperscale profile.
(cherry picked from commit c41c4b2bf3)
Instead of including distribution specific files in the subimages,
let's have one common mkosi.pkgenv/ directory that handles all the
matching which is then included in the subimages.
This gives us more control on exactly how we do the matching.
(cherry picked from commit fb514c2f8f)
