shell_escape() is mostly used for mount paths and similar, where we assume
no newlines are present in the string. But if any were ever present, we
should escape them. So let's simplify the code by making this unconditional.
We had the following scenario:
under /etc/systemd/system/
- foo@.service
- bar@tty12.service → foo@tty12.service
- multi-user.target.wants/foo@tty12.service
Existing code did not "know" that foo@tty12.service has alias bar@tty12.service:
$ systemctl show -P Names foo@tty12.service
foo@tty12.service
Since multi-user.target is always loaded, we would load foo@tty12.service.
When trying to load bar@tty12.service, it would (correctly) detect that
bar@tty12.service is an alias for foo@tty12.service, and try to merge the
bar@tty12.service unit into the foo@tty12.service. This would fail, because
foo@tty12.service was already loaded, and only about-to-be-loaded units can
be merged.
With the patch we consider bar@tty12.service an alias of foo@tty12.service
immediately, so the issue does not occur:
$ systemctl show -P Names foo@tty12.service
foo@tty12.service bar@tty12.service
Fixes#19409.
This turned in a bigger rewrite. The logic add "the main name and all aliases"
was implemented twice, slightly different in both cases. I split that part out
to a new function. The result about the same length, but hopefully a bit easier
to read.
Logging output is also improved a bit. Some left-over debug logs have been
removed or cleaned up.
This is a fairly big change, but (with the addition in the following commit),
we have pretty good coverage of this logic.
New glibc deprecated mallocinfo(), even newer glibc added mallocinfo2()
as replacement. Use it, if it exists.
Follow-up for 4b6f74f5a0 and related
commits.
We would try to return a value that could be nonzero only if the kernel
reported writing more bytes than we gave to it, hopefully a rare occurence.
Instead, assert that this doesn't happen.
Instead, return true if we got to the end of the iovec array. The caller
can use this information to know that the whole iovec array was written.
This allows one loop to be dropped in write_to_syslog().
Also drop _unlikely_: this function is called with very short arrays, and
it *is* likely that we trigger this condition. Let's just let the compiler
generate normal code without giving it a potentially false hint.
DMI vendor information fields do not provide enough information for us to
distinguish between Amazon EC2 virtual machines and bare-metal instances.
SMBIOS provides a BIOS Information
table (https://www.dmtf.org/sites/default/files/standards/documents/DSP0134_3.4.0.pdf
Ch. 7) that provides a field to indicate that the current machine is a virtual
machine. On EC2 virtual machine instances, this field is set, while bare-metal
instances leave this unset, so we inspect the field via the kernel's
/sys/firemware/dmi/entries interface.
Fixes#18929
Hidden and backup files cannot be valid unit name (we reject anything
starting with a dot, and we require type suffixes). So let's not iterate
over those at all.
The ifdef pattern is the same for all syscalls, so most of the time, if one is
not defined, all others will too. So let's reduce the noise a bit and emit one
warning in case the support for the architecture is fully missing. (Current
template was copied over from before when we added numbers for each syscall by
hand and stopped making sense when we started generating the header from a
table that is expected to have all syscall numbers.)
I found myself often looking for a quick way to determine "the local IP
address", and then being lost in the "ip addr" output to find for the
right one to use. This is supposed to help a bit with that. Let's
introduce a new special hostname "_outbound" with semantics similar to
"_gateway" that resolves to addresses that are the closest I could come
up with that maps to "the" local IP address.
On some conditions (particularly when mobile CPUs are going to sleep),
the posix_fallocate(), which is called when a new journal file is allocated,
can return -1 (EINTR). This is counted as a fatal error. So the journald
closes both old and journals, and simply throwing away further incoming
events, because of no log files open.
Introduce posix_fallocate_loop() that restarts the function in the case
of EINTR. Also let's make code base more uniform by returning negative
values on error.
Fix assert in test-sigbus.c that incorrectly counted positive values as
success. After changing the function return values, that will actually work.
Fixes: #19041
Signed-off-by: Igor Zhbanov <i.zhbanov@omprussia.ru>
So far all file systems where checked by instances of
systemd-fsck@.service, with the exception of the root fs which was
covered by systemd-fsck-root.service. The special handling is necessary
to deal with ordering issues: we typically want the root fs to be
checked before all others, and — weirdly — allow mounting it before the
fsck done (for compat with initrd-less boots).
This adds similar special handling for /usr: if the hierarchy is placed
on a separate file system check it with a special
systemd-fsck-usr.service instead of a regular sysemd-fsck@.service
instance. Reason is again ordering: we want to allow mounting of /usr
without the root fs already being around in the initrd, to cover for
cases where the root fs is created on first boot and thus cannot be
mounted/checked before /usr.
This changes the fstab-generator to handle mounting of /usr/ a bit
differently than before. Instead of immediately mounting the fs to
/sysroot/usr/ we'll first mount it to /sysusr/usr/ and then add a
separate bind mount that mounts it from /sysusr/usr/ to /sysroot/usr/.
This way we can access /usr independently of the root fs, without for
waiting to be mounted via the /sysusr/ hierarchy. This is useful for
invoking systemd-repart while a root fs doesn't exist yet and for
creating it, with partition data read from the /usr/ hierarchy.
This introduces a new generic target initrd-usr-fs.target that may be
used to generically order services against /sysusr/ to become available.
This fixes maybe-uninitialized warning:
```
../src/basic/fileio.c: In function ‘chase_symlinks_and_fopen_unlocked’:
../src/basic/fileio.c:1026:19: warning: ‘f’ may be used uninitialized in this function [-Wmaybe-uninitialized]
1026 | *ret_file = f;
| ~~~~~~~~~~^~~
```
As @yuwata correctly points out, this became broken when log_debug()
started returning -EIO. I wanted to preserve this pattern, but it turns
out it is not very widely used, and preserving it would make the whole
thing, already quite complicated, even more complex.
log_debug() is made like log_info() and friends, and returns void.
Let's assert if we ever happen to pass 0 to one of the log functions.
With the preceding commit to return -EIO from log_*(), passing 0 wouldn't
affect the return value any more, but it is still most likely an error.
The unit test code is an exception: we fairly often pass the return value
to print it, before checking what it is. So let's assert that we're not
passing 0 in non-test code. As with the previous check for %m, this is only
done in developer mode. We are depending on external code setting
errno correctly for us, which might not always be true, and which we can't
test, so we shouldn't assert, but just handle this gracefully.
I did a bunch of greps to try to figure out if there are any places where
we're passing 0 on purpose, and couldn't find any.
The one place that failed in tests is adjusted.
About "zerook" in the name: I wanted the suffix to be unambiguous. It's a
single "word" because each of the words in log_full_errno is also meaningful,
and having one term use two words would be confusing.
This is only done in developer mode. It is a pretty rare occurence that we
make this kind of mistake. And even if it happens, the result is just a misleading
error message. So let's only do the check in non-release builds.
This silences some warnigns where gcc thinks that some variables are
unitialized. One particular case:
../src/journal/journald-server.c: In function 'ache_space_refresh':
../src/journal/journald-server.c:136:28: error: 'vfs_avail' may be used uninitialized in this function [-Werror=maybe-uninitialized]
136 | uint64_t vfs_used, vfs_avail, avail;
| ^~~~~~~~~
../src/journal/journald-server.c:136:18: error: 'vfs_used' may be used uninitialized in this function [-Werror=maybe-uninitialized]
136 | uint64_t vfs_used, vfs_avail, avail;
| ^~~~~~~~
cc1: all warnings being treated as errors
which is caused by
d = opendir(path);
if (!d)
return log_full_errno(errno == ENOENT ? LOG_DEBUG : LOG_ERR,
errno, "Failed to open %s: %m", path);
if (fstatvfs(dirfd(d), &ss) < 0)
return log_error_errno(errno, "Failed to fstatvfs(%s): %m", path);
For some reason on aarch64 gcc thinks we might return non-negative here. In
principle errno must be set in both cases, but it's hard to say for certain.
So let's make sure that our code flow is correct, even if somebody forgot to
set the global variable somewhere.
Using a enum is all nice and generic, but at this point it seems unlikely that
we'll add further build modes. But having an enum means that we need to include
the header file with the enumeration whenerever the conditional is used. I want
to use the conditional in log.h, which makes it hard to avoid circular imports.
With some versions of the compiler, the _cleanup_ attr makes it think
the variable might be freed/closed when uninitialized, even though it
cannot happen. The added cost is small enough to be worth the benefit,
and optimized builds will help reduce it even further.
Let's define both an enum and a typedef named SpecialGlyph, the way we
usually do it.
Also, introduce an "invalid" special glyph, assigned to -EINVAL, also
like we always do it. (And handle it somewhat sanely in special_glyph()
