This reverts commit 097537f07a.
At least Fedora and Debian have already reverted this at the distro
level because it causes more problems than it solves. Arch is debating
reverting it as well [0] but would strongly prefer that this happens
upstream first. Fixes#15188.
[0] https://bugs.archlinux.org/task/66458
Make use of curl_multi_assign to associate each IO sd_event_source with
a CURL object. This means we always get passed the right event source
and don't need to worry about looking up the associated CURL object,
particularly in the case where the FD has been closed on a REMOVE event.
After curl 7.20.0, this function never returns negative error codes.
Make this consistent with the other call to this function and only
compare against CURLM_OK.
Prior to this commit, a .link file with a [Match] section containing
MACAddress= would match any device without a MAC. This restores the
matching logic prior to e90d037.
Since a581e45ae8, there's a few function calls to
unit_new_for_name which will unit_free on failure. Prior to this commit,
a failure would result in calling unit_free with a NULL unit, and hit an
assertion failure, seen at least via device_setup_unit:
Assertion 'u' failed at src/core/unit.c:519, function unit_free(). Aborting.
Fixes#4747https://bugs.archlinux.org/task/51950
Delete the dbus1 generator and some critical wiring. This prevents
kdbus from being loaded or detected. As such, it will never be used,
even if the user still has a useful kdbus module loaded on their system.
Sort of fixes#3480. Not really, but it's better than the current state.
This fixes 2 problems introduced by 6feeeab0bc:
1) If name_to_handle_at returns ENOSYS for the child, we'll wrongly
return -ENOSYS when it returns the same for the parent. Immediately
jump to the fallback logic when we get ENOSYS.
2) If name_to_handle_at returns EOPNOTSUPP for the child but suceeds
for the parent, we'll be comparing an uninitialized value (mount_id) to
an initialized value (mount_id_parent). Initialize the mount_id
variables to invalid mount_ids to avoid this.
The code already calls sd_notify("READY=1"), so we may as well take
advantage of the startup behavior in the unit. The same was done for
the journal in a87a38c20.
f7101b7368 copied some logic to prevent enabling masked units, but
also added a check which causes attempts to enable templated units to
fail. Since we know the logic beyond this check will properly handle
units which truly do not exist, we can rely on the unit file state
comparison to suffice for expressing the intent of f7101b7368.
ref: https://bugs.archlinux.org/task/42616
A combination of commits f3c80515c and 79d80fc14 cause nspawn to
silently fail with a commandline such as:
# systemd-nspawn -D /build/extra-x86_64 --bind=/usr
strace shows the culprit:
[pid 27868] writev(2, [{"Failed to create mount point /build/extra-x86_64/usr: File exists", 82}, {"\n", 1}], 2) = 83
We already call this on via bus_event_loop_with_idle on exit. This
makes machined consistent with other similar daemons: localed,
hostnamed, timedated.
Since c6a373a263, we might encounter unit templates via the
'list-units' verb. These aren't restartable (and we throw errors), so
make sure they're filtered out of the completion options.
fixes downstream bug: https://bugs.archlinux.org/task/41719
Cases where name_to_handle_at is used allocated the full struct to be
MAX_HANDLE_SZ, and assigned this size to handle_bytes. This is wrong
since handle_bytes should describe the length of the flexible array
member and not the whole struct.
Define a union type which includes sufficient padding to allow
assignment of MAX_HANDLE_SZ to be correct.
Since the index is already post-incremented when the array is appended
to, this assertion can be wrongly reached when the array is at capacity
(with the NULL terminator). The bug is reproducible on shutdown with
the following settings in /etc/systemd/system.conf:
LogTarget=journal-or-kmsg
LogColor=yes
LogLocation=yes
Reported by Thermi on IRC.
Prior to 3.2, /proc/sys/kernel/hostname isn't a pollable file and
sd_event_add_io will return EPERM. Ignore this failure, since it isn't
critical to journald operation.
Reported and tested by user sraue on IRC.
Arch Linux uses nspawn as a container for building packages and needs
to be able to start a 32bit chroot from a 64bit host. 24fb111207
disrupted this feature when seccomp handling was added.
The behavior of this is a little cryptic in that $MAINPID must exit as
a direct result of receiving a signal in order for a listed signal to
be considered a success condition.
Since this is a shortopt with an optional argument, assume the user
knows what they're doing. The longopts --boot and --this-boot will
continue to offer boot IDs as completions.
This flag shouldn't try and consume the following argument. It should
behave like every other flag which takes an optional argument when
parsed by getopt_long.
Since 0c6f1f4ea4 the check was useless, because the kernel will
ever only write "partition" or "file" there.
OTOH, it is possible that "\\040(deleted)" (escaped " (deleted)")
will be added for removed files. This should not happen, so add
a warning to detect those cases.
Previously, if a file's bind mount destination didn't exist, nspawn
would blindly create a directory, and the subsequent bind mount would
fail. Examine the filetype of the source and ensure that, if the
destination does not exist, that it is created appropriately.
Also go one step further and ensure that the filetypes of the source
and destination match.
In the process, rename udev_encode_string which is poorly named for what
it does. It deals specifically with encoding names that udev creates and
has its own rules: utf8 is valid but some ascii is not (e.g. path
separators), and everything else is simply escaped. Rename it to
encode_devnode_name.
There's now some more obvious overlap amongst the two utf8 validation
functions, but no more than there already was previously.
This also adds some menial tests for anyone who wants to do more
merging of these two in the future.
The AA is unnecessary and only adds needless complexity. Replace it
with a case statement instead of repeatedly calling __contains_word to
overglorify string equalities.
- scope the iterator var
- use the correct, quoted, non-expansion prone positional parameter
notation
- prevent expansion on RHS of comparison
- remove unneeded explicit returns.
This really should be defined only once...
PKG_CHECK_EXISTS won't created a cached variable that later messes with
our PKG_CHECK_MODULES check for an explicit version. Unfortunately,
nesting these checks as the code existed lead to an odd error. Rather,
split the checks apart.
This also improves to the error message when the requisite version
isn't found, and supplies the literal version systemd needs.
This doesn't need to be passed, as it's handled by libtool. Since the
default for autoconf is --disable-static, this change is effectively a
noop. It only matters if you pass --enable-static, in which case the
static libs for systemd libraries will actually be built.
Nitpicky, but this only affects systemd libs. The override for the
other libs remains since these libs are always loaded dynamically and
never compiled staticly.
Dave Reisner
Dave Reisner