Merge members mask invalidation into
unit_add_siblings_to_cgroup_realize_queue, this way unit_realize_cgroup
needn't be called with members mask invalidation.
We have to retain the members mask invalidation in unit_load -- although
active units would have cgroups (re)realized (unit_load queues for
realization), the realization would happen with potentially stale mask.
unit_free(u) realizes direct parent and invalidates members mask of all
ancestors. This isn't sufficient in v1 controller hierarchies since
siblings of the freed unit may have existed only because of the removed
unit.
We cannot be lazy about the siblings because if parent(u) is also
removed, it'd migrate and rmdir cgroups for siblings(u). However,
realized masks of siblings(u) won't reflect this change.
This was a non-issue earlier, because we weren't removing cgroup
directories properly (effectively matching the stale realized mask),
removal failed because of tasks left by missing migration (see previous
commit).
Therefore, ensure realization of all units necessary to clean up after
the free'd unit.
Fixes: #14149
When we are about to derealize a controller on v1 cgroup, we first
attempt to delete the controller cgroup and migrate afterwards. This
doesn't work in practice because populated cgroup cannot be deleted.
Furthermore, we leave out slices from migration completely, so
(un)setting a control value on them won't realize their controller
cgroup.
Rework actual realization, unit_create_cgroup() becomes
unit_update_cgroup() and make sure that controller hierarchies are
reduced when given controller cgroup ceased to be needed.
Note that with this we introduce slight deviation between v1 and v2 code
-- when a descendant unit turns off a delegated controller, we attempt
to disable it in ancestor slices. On v2 this may fail (kernel enforced,
because of child cgroups using the controller), on v1 we'll migrate
whole subtree and trim the subhierachy. (Previously, we wouldn't take
away delegated controller, however, derealization was broken anyway.)
Fixes: #14149
When we're disabling controller on a direct child of root cgroup, we
forgot to add root slice into cgroup realization queue, which prevented
proper disabling of the controller (on unified hierarchy).
The mechanism relying on "bounce from bottom and propagate up" in
unit_create_cgroup doesn't work on unified hierarchy (leaves needn't be
enabled). Drop it as we rely on the ancestors to be queued -- that's now
intentional but was artifact of combining the two patches:
cb5e3bc37d ("cgroup: Don't explicitly check for member in UNIT_BEFORE") v240~78
65f6b6bdcb ("core: fix re-realization of cgroup siblings") v245-rc1~153^2
Fixes: #14917
The current implementation is LIFO, which is a) confusing b) prevents
some ordered operations on the cgroup tree (e.g. removing children
before parents).
Fix it quickly. Current list implementation turns this from O(1) to O(n)
operation. Rework the lists later.
When a prefix is delegated to an interface that is already sending
RAs, send an RA immediately to inform clients of the new prefix.
This allows them to start using it immediately instead of waiting
up to nearly 10 minutes (depending on when the last timed RA was
sent). This type of situation might occur if, for example, an
outage of the WAN connection caused the addresses and prefixes to
be lost and later regained after service was restored. The
condition for the number of RAs sent being above 0 simultaneously
ensures that RADV is already running and that this code doesn't
send any RAs before the timed RAs have started when the interface
first comes up.
Let's find the right os-release file on the host side, and only mount
the one that matters, i.e. /etc/os-release if it exists and
/usr/lib/os-release otherwise. Use the fixed path /run/host/os-release
for that.
Let's also mount /run/host as a bind mount on itself before we set up
/run/host, and let's mount it MS_RDONLY after we are done, so that it
remains immutable as a whole.
It needs to be world readable (unlike /etc/shadow) when created anew.
This fixes systems that boot with "systemd-nspawn --volatile=yes", i.e.
come up with an entirely empty /etc/ and thus no existing /etc/passwd
file when firstboot runs.
We want our OS trees to be MS_SHARED by default, so that our service
namespacing logic can work correctly. Thus in nspawn we mount everything
MS_SHARED when organizing our tree. We do this early on, before changing
the user namespace (if that's requested). However CLONE_NEWUSER actually
resets MS_SHARED to MS_SLAVE for all mounts (so that less privileged
environments can't affect the more privileged ones). Hence, when
invoking it we have to reset things to MS_SHARED afterwards again. This
won't reestablish propagation, but it will make sure we get a new set of
mount peer groups everywhere that then are honoured for the mount
namespaces/propagated mounts set up inside the container further down.
[zjs: Looking at https://packages.ubuntu.com/bionic/iptables-dev, iptables-dev
was a transitional package that was pulling in libxtables-dev, libip4tc-dev,
and libip6tc-dev (as listed by @GiedriusS). iptables-dev is gone in focal, so
replace it by the expanded list.]
This API is a complete mess. We forgot to do a hashed comparison for duplicate
entries and we use a direct pointer comparison. For trivial_hash_ops the result
is the same. For all other case, it's not. Fixing this properly will require
auditing all the uses of set_put() and ordered_set_put(). For now, let's just
acknowledge the breakage.
Currently, each change to NEWS triggers a meson reconfigure that
changes SOURCE_EPOCH which causes a full rebuild. Since NEWS changes
relatively often, we have a full rebuild each time we pull from
master even if we pull semi-regularly. This is further compounded
when using branches since NEWS has a relatively high chance to
differ between branches which causes git to update the modification
time, leading to a full rebuild when switching between branches.
We fix this by using the creation time of the latest git tag instead.
There are a lot of edge cases that the current implementation
doesn't handle, especially in cases where one of passwd/shadow
exists and the other doesn't exist. For example, if
--root-password is specified, we will write /etc/shadow but
won't add a root entry to /etc/passwd if there is none.
To fix some of these issues, we constrain systemd-firstboot to
only modify /etc/passwd and /etc/shadow if both do not exist
already (or --force) is specified. On top of that, we calculate
all necessary information for both passwd and shadow upfront so
we can take it all into account when writing the actual files.
If no root password options are given --force is specified or both
files do not exist, we lock the root account for security purposes.
Fixes#16401.
c80a9a33d0 introduced the .can_fail field,
but didn't set it on .targets. Targets can fail through dependencies.
This leaves .slice and .device units as the types that cannot fail.
$ systemctl cat bad.service bad.target bad-fallback.service
[Service]
Type=oneshot
ExecStart=false
[Unit]
OnFailure=bad-fallback.service
[Service]
Type=oneshot
ExecStart=echo Fixing everythign!
$ sudo systemctl start bad.target
systemd[1]: Starting bad.service...
systemd[1]: bad.service: Main process exited, code=exited, status=1/FAILURE
systemd[1]: bad.service: Failed with result 'exit-code'.
systemd[1]: Failed to start bad.service.
systemd[1]: Dependency failed for bad.target.
systemd[1]: bad.target: Job bad.target/start failed with result 'dependency'.
systemd[1]: bad.target: Triggering OnFailure= dependencies.
systemd[1]: Starting bad-fallback.service...
echo[46901]: Fixing everythign!
systemd[1]: bad-fallback.service: Succeeded.
systemd[1]: Finished bad-fallback.service.
This reverts commit c7220ca802.
The removal was done as a reaction to the messages from systemd:
initrd-root-fs.target: Requested dependency OnFailure=emergency.target ignored (target units cannot fail).
initrd.target: Requested dependency OnFailure=emergency.target ignored (target units cannot fail).
initrd-root-device.target: Requested dependency OnFailure=emergency.target ignored (target units cannot fail).
initrd-fs.target: Requested dependency OnFailure=emergency.target ignored (target units cannot fail).
local-fs.target: Requested dependency OnFailure=emergency.target ignored (target units cannot fail).
...
But it seems that the messages themselves are wrong, and the units were OK.
The Address objects in the set generated by ndisc_router_generate_addresses()
have the equivalent prefixlen, flags, prefered lifetime.
This commit makes ndisc_router_generate_addresses() return Set of
in6_addr.
Michal Koutný
Daan De Meyer
Michael Marley
Christian Göttsche
Piotr Drąg
Yu Watanabe
Axel Rasmussen
Dan Streetman
Lennart Poettering
Luca Boccassi
Zbigniew Jędrzejewski-Szmek
szb512