- Drop fstat() fallback path now that we assume fdinfo
is available
- Use at_flags_normalize_nofollow()
- Accept empty path the same way as NULL
- Accept fd being AT_FDCWD and filename being "."
This doesn't make the RNG cryptographic strength, but if we have it
easily accessible, why not include the pidfd id. It is after all not
vulnerable to reuse.
Note that this drops a lot of "const" qualifiers on PidRef arguments.
That's because pidref_is_self() suddenly might end changing the PidRef
because it acquires the pidfd ID.
We had this previously already with pidfd_equal(), but this amplifies
the problem.
I guess we C's "const" doesn't really work for stuff that contains
caches, that is just conceptually constant, but not actually.
I just wanted to switch the machine id setup code to use chase() or its
changes, given it supports --root=/--image= operation. That turned out
to be a rabbit hole, and became much bigger...
let's move final processing of the filename out of the loop, and apply
it in all cases, uniformly, even if we are asked to only return the
final filename.
This also implies the new CHASE_MUST_BE_DIRECTORY flag in case the
specified path ends in a slash. This makes the rules stricter, it means
we'll be closer to how this is handled in kernel: if a path ends in a
slash it can never refer to a non-directory.
This allows using CHASE_MKDIR_0755 without CHASE_NONEXISTENT or
CHASE_PARENT, so that it will create the final component of the path
too should it be missing.
This is really useful as a mkdir_p() replacement that returns an fd to
the final component, and knows how to operate relative to a root fs.
Kinda reverts 4ea0bcb922 (which only
refused the flags combination which didn't work, instead of making it
work, which is what this commit does.)
This also corrects behaviour if CHASE_MKDIR_0755 is used in one more
way: we'll now always open the dir as O_PATH. This is generally the
better idea, but matters in particular once with allow using
CHASE_MKDIR_0755 to create the final component: we should uniformly
return an O_PATH dir that must be converted to a proper fd first before
using it.
Let's give the user control on how to handle JSON "null" assignments of
the log level. As one of three cases: as failure, as LOG_NULL (i.e. to
turn off logging) or as LOG_INFO (as our usual default log level).
Let's then use that in the generic SetLogLevel() call, so that callers
can use it to explicitly turn off logging in a service.
Note that this was (probably accidentally) already implemented, except
that the introspection enforcement blocked it. Let's clean this up and
make this officially a thing, since it's generally useful to turn off
logging I think.
Often we want to fork off a process that just hangs until we kill it,
let's add a simple flag to create one of this type, and use it at
various places.
Fixes the following error:
```
../src/basic/random-util.c: In function "fallback_random_bytes":
../src/basic/random-util.c:45:26: error: initializer-string for array of "char" is too long [-Werror=unterminated-string-initialization]
45 | .label = "systemd fallback random bytes v1",
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
```
Unfortunately kernel reports EOF if there's an inconsistency between efivarfs var list
and what's actually stored in firmware, c.f. #34304. A zero size env var is not allowed in
efi and hence the variable doesn't really exist in the backing store as long as it is zero
sized, and the kernel calls this "uncommitted". Hence we translate EOF back to ENOENT here,
as with kernel behavior before
https://github.com/torvalds/linux/commit/3fab70c165795431f00ddf9be8b84ddd07bd1f8f
If the kernel changes behaviour (to flush dentries on resume), we can drop
this at some point in the future. But note that the commit is 11
years old at this point so we'll need to deal with the current behaviour for
a long time.
Fix#34304.
in strv_new() we have STRV_IGNORE for skipping over an argument in the
argument list. Let's add the same to strextend_with_separator():
strextend_with_separator(&x, "foo", POINTER_MAX, "bar");
will result in "foobar" being appended to "x". (POINTER_MAX Which is
different from NULL, which terminates the argument list).
This is useful for ternary op situations.
(We should probably get rid of STRV_IGNORE and just use POINTER_MAX
everywhere directly, but that's for another time.)
The functions are very similar, let's make them the same. If the first
argument to strextend() is NULL instead of extending a string we'll
allocate a fresh one and return that.
Background: Fedora/RHEL are switching to sysusers.d metadata for
creation of users and groups for system users defined by packages
(https://fedoraproject.org/wiki/Changes/RPMSuportForSystemdSysusers).
Packages carry sysusers files. During package installation, rpm calls an
program to execute on this config. This program may either be
/usr/lib/rpm/sysusers.sh which calls useradd/groupadd, or
/usr/bin/systemd-sysusers. To match the functionality provided by
useradd/groupadd from the shadow-utils project, systemd-sysusers must
emit audit events so that it provides a drop-in replacement.
systemd-sysuers will emit audit events AUDIT_ADD_USER/AUDIT_ADD_GROUP
when adding users and groups. The operation "names" are copied from
shadow-utils, so the format of the events that is generated on success
should be identical. On failure, things are more complicated. We write
the whole file at once, once, so we first generate "success" messages
for each entry, then we try to write the files, and if things fail, we
generate failure messages to all entries that we failed to write.
Background: Fedora/RHEL are switching to sysusers.d metadata for creation of
users and groups for system users defined by packages
(https://fedoraproject.org/wiki/Changes/RPMSuportForSystemdSysusers).
Packages carry sysusers files. During package installation, rpm calls an
program to execute on this config. This program may either be
/usr/lib/rpm/sysusers.sh which calls useradd/groupadd, or
/usr/bin/systemd-sysusers. To match the functionality provided by
useradd/groupadd from the shadow-utils project, systemd-sysusers must emit
audit events so that it provides a drop-in replacement.
systemd-sysuers will emit audit events AUDIT_ADD_USER/AUDIT_ADD_GROUP when
adding users and groups. The operation "names" are copied from shadow-utils in
Fedora (which has a patch to change them from the upstream version), so the
format of the events that is generated on success should be identical.
The helper code is shared between sysusers and utmp-wtmp. I changed the
audit_fd variable to be unconditional. This way we can avoid ugly iffdefery
every time the variable would be used. The cost is that 4 bytes of unused
storage might be present. This is negligible, and the compiler might even be
able to optimize that away if it inlines things.
Lennart Poettering
Mike Yuan
Yu Watanabe
hanjinpeng
wrvsrx
Zbigniew Jędrzejewski-Szmek