Its value depends on how strerror_r() implemented, and the
implementations of the function in glibc and musl are actually
different. Let's drop the checks.
musl provides XSI compliant strerror_r(), and it is slightly different
from the one by glibc.
Let's introduce a tiny wrapper to convert XSI strerror_r() to GNU one.
The wrapper also patches musl's spurious catchall error message.
glibc's sys/stat.h includes linux/stat.h, and we have copy of it from
the latest kernel, hence all new flags are always defined.
However, musl's sys/stat.h does not include linux/stat.h, and moreover,
they conflict with each other, hence we cannot include both header
simultaneously. Let's define missing macros to support musl.
Even 'gshadow' meson option is disabled, src/shared/userdb.c and
src/shared/user-record-nss.c include gshadow.h unconditionally.
Let's introduce dummy header to make them compiled gracefully.
musl's CPU_ISSET_S() macro does not avoid multiple evaluations, and it
only accepts simple variable or constant.
Fixes the following error.
```
../src/shared/cpu-set-util.c: In function ‘cpu_set_to_mask_string’:
../src/shared/cpu-set-util.c:101:41: warning: operation on ‘i’ may be undefined [-Werror=sequence-point]
101 | if (CPU_ISSET_S(--i, c->allocated, c->set))
| ^
```
glibc defines HOST_NAME_MAX as 64 and our code rely on that, but musl
defines the constant as 255. Let's provide our own definition for the
maximum length.
glibc defines AT_FDCWD as -100, but musl defines it as (-100).
In basic-forward.h, we also define AT_FDCWD as -100, hence musl's fcntl.h
conflicts with forward.h. This is for avoiding the conflict.
- musl does not support gshadow, and does not provide gshadow.h,
- musl does not support nss, and does not provide nss.h which is necessary
for each nss modules,
- musl does not provide NI_IDN.
Otherwise, musl defines wchar_t as int, which conflicts with the
assumption by sd-boot, i.e. wchar_t is 2 bytes.
Fixes the following build error:
```
In file included from ../src/boot/efi-log.h:4,
from ../src/boot/linux_x86.c:13:
../src/boot/efi.h:19:24: error: conflicting types for 'wchar_t'; have 'short unsigned int'
19 | typedef __WCHAR_TYPE__ wchar_t;
| ^~~~~~~
In file included from /usr/include/stddef.h:19,
from ../src/boot/efi.h:9:
/usr/include/bits/alltypes.h:10:13: note: previous declaration of 'wchar_t' with type 'wchar_t' {aka 'int'}
10 | typedef int wchar_t;
| ^~~~~~~
```
glibc sets it when _GNU_SOURCE is defined, however, musl does not.
Let's explicitly define it to make getdents64() and struct dirent64
available even when building with musl.
musl does not set tm_wday when it is explicitly requested.
The check is not necessary at all, it is just for safety.
Let's skip it when built with musl.
musl v1.2.5 does not support %z specifier in strptime(). Since
https://github.com/kraj/musl/commit/fced99e93daeefb0192fd16304f978d4401d1d77
%z is supported, but it only supports strict RFC-822/ISO 8601 format,
that is, 4 digits with sign (e.g. +0900 or -1400), but does not support
extended format: 2 digits or colon separated 4 digits (e.g. +09 or -14:00).
Let's add fallback logic to make it support the extended timezone spec.
sys/prctl.h anyway includes linux/prctl.h and actually these .c files
includes sys/prctl.h. Hence, it is not necessary to explicitly include
linux/prctl.h.
musl's newlocale() always provides a locale object even the requested
locale does not exist. Let's explicitly check the existence of the
requested locale file.
The --global flag has been broken since commit
d77d42ed3a, which added a
blanket restriction on acquiring D-Bus connections when
arg_runtime_scope is RUNTIME_SCOPE_GLOBAL. This was done to prevent
crashes, but inadvertently broke legitimate use cases like 'systemctl
edit --global' and 'systemctl cat --global'.
The issue is that verb_edit() and verb_cat() were unconditionally
calling acquire_bus(), which triggers the restriction and fails with
"--global is not supported for this operation."
This commit fixes the issue by making bus acquisition conditional,
following the same pattern used in verb_enable():
- Only acquire the bus when install_client_side() returns NO (i.e., for
system and user scopes)
- For client-side operations (--global, --root, etc.), skip bus
acquisition and use mangle_names() instead of expand_unit_names()
- Update find_paths_to_edit() and verb_cat() to handle NULL bus by
forcing client-side path lookups
- Skip bus-dependent checks (unit_is_masked, need_daemon_reload) when
bus is NULL
This allows both 'systemctl edit --global' and 'systemctl cat --global'
to work correctly by performing all operations client-side without
requiring a connection to the system or user manager.
Fixes#31272
Aim of this patches set, is to add a new type SD_PATH_SEARCH_SYSCTL for
sd_path_lookup() and sd_path_lookup_strv(). This new type is used to get the
directories list used by systemd-sysctl:
- /etc/sysctl.d/
- /run/sysctl.d/
- /usr/local/lib/sysctl.d/
- /usr/lib/sysctl.d/
This implements the change in libsystemd, systemd-path, and systemd-sysctl.
The same service name was accidentally used for two invocations:
```
[ 1801.197993] H TEST-04-JOURNAL.sh[20563]: + assert_rc 0 journalctl -q -D /run/log/journal/e30adae55e664d328af442bf5df694c8/ -u test-23833.service --grep service=test-23833.service
[ 1801.198527] H TEST-04-JOURNAL.sh[20685]: + set +ex
[ 1801.222676] H TEST-04-JOURNAL.sh[20686]: Nov 10 03:18:51 H systemd[1]: test-23833.service: About to execute: /usr/bin/bash -c "echo service=test-23833.service invocation=\$INVOCATION_ID; journalctl --sync"
[ 1801.222676] H TEST-04-JOURNAL.sh[20686]: Nov 10 03:18:51 H systemd[1]: Started test-23833.service - [systemd-run] /usr/bin/bash -c "echo service=test-23833.service invocation=\$INVOCATION_ID; journalctl --sync".
[ 1801.222676] H TEST-04-JOURNAL.sh[20686]: Nov 10 03:18:51 H (bash)[20681]: test-23833.service: Executing: /usr/bin/bash -c "echo service=test-23833.service invocation=\$INVOCATION_ID; journalctl --sync"
[ 1801.222676] H TEST-04-JOURNAL.sh[20686]: Nov 10 03:18:51 H bash[20681]: service=test-23833.service invocation=1866f15e95924a688dcecde72bf345f6
[ 1801.227878] H TEST-04-JOURNAL.sh[20563]: + assert_rc 1 journalctl -q -D /var/log/journal/e30adae55e664d328af442bf5df694c8/ -u test-23833.service --grep service=test-23833.service
[ 1801.228265] H TEST-04-JOURNAL.sh[20689]: + set +ex
[ 1801.253412] H TEST-04-JOURNAL.sh[20690]: Nov 10 03:18:49 H systemd[1]: test-23833.service: About to execute: /usr/bin/bash -c "echo service=test-23833.service invocation=\$INVOCATION_ID; journalctl --sync"
[ 1801.253412] H TEST-04-JOURNAL.sh[20690]: Nov 10 03:18:49 H systemd[1]: Started test-23833.service - [systemd-run] /usr/bin/bash -c "echo service=test-23833.service invocation=\$INVOCATION_ID; journalctl --sync".
[ 1801.253412] H TEST-04-JOURNAL.sh[20690]: Nov 10 03:18:49 H (bash)[20581]: test-23833.service: Executing: /usr/bin/bash -c "echo service=test-23833.service invocation=\$INVOCATION_ID; journalctl --sync"
[ 1801.253412] H TEST-04-JOURNAL.sh[20690]: Nov 10 03:18:49 H bash[20581]: service=test-23833.service invocation=a3089a62b5624d21bac0a75a3995d8b5
[ 1801.258158] H TEST-04-JOURNAL.sh[20692]: FAIL: expected: '1' actual: '0'
```
systemd-repart is incorrectly choosing the loop-mount
code path to copy files after formatting, instead of using the --rootdir
path, which is required by mkfs.btrfs to apply compression (since it's
on files, not the fs).
So two fixes (and an integ test):
1. If Btrfs compression is requested without a root directory (e.g.,
Compression= without CopyFiles=), we now log a warning and skip the
--compress flag. This prevents the mkfs.btrfs failure, and it's
meaningless anyway without any files.
2. The logic in repart now uses the --rootdir code path whenever the
partition is btrfs and compression is requested. Otherwise it still
won't work even in the legitimate case because use the loop mounting
code, which is too late to use --compress.
Fixes: https://github.com/systemd/systemd/issues/39584
- use loop for checking existence of functions,
- rename HAVE_LIBARCHIVE_XYZ -> HAVE_ARCHIVE_XYZ to make them match with
the function name,
- do not conditionally include user-util.h in libarchive-util.h,
- sort library function symbols.
Fixes#38486 again, which was fixed by
1405d46bf9, but regressed again by
f875a8026e.
This also make it use exec_input_is_terminal().
Follow-up for f875a8026e.
Before aa47d8ade1, when we failed to lock
the device node, we simply trigger change events for the device and its
partitions. But the commit killed the fallback logic. Let's restore that.
Fixes#39544.
Yu Watanabe
