Reproducer:
$ meson build && cd build
$ ninja
$ sudo useradd test
$ sudo su test
$ ./systemd --system --test
...
Failed to create /user.slice/user-1000.slice/session-6.scope/init.scope control group: Permission denied
Failed to allocate manager object: Permission denied
Above error message is caused by the fact that user test didn't have its
own session and we tried to set up init.scope already running as user
test in the directory owned by different user.
Let's skip setting up init.scope altogether since we won't be launching
processes anyway.
We maintain a queue of units and jobs that we are supposed to generate
change/new notifications for because they were either just created or
some of their property has changed. Let's throttle processing of this
queue a bit: as soon as > 1K of bus messages are queued for writing
let's skip processing the queue, and then recheck on the next
iteration again.
Moreover, never process more than 100 units in one go, return to the
event loop after that. Both limits together should put effective limits
on both space and time usage of the function, delaying further
operations until a later moment, when the queue is empty or the the
event loop is sufficiently idle again.
This should keep the number of generated messages much lower than
before on busy systems or where some client is hanging.
Note that this also means a bad client can slow down message dispatching
substantially for up to 90s if it likes to, for all clients. But that
should be acceptable as we only allow trusted bus clients, anyway.
Fixes: #8166
config_parse_join_controllers would free the destination argument on failure,
which is contrary to our normal style, where failed parsing has no effect.
Moving it to shared also allows a test to be added.
A .socket will reference a .service unit, by registering a UnitRef with the
.service unit. If this .service unit has the .socket unit listed in Wants or
Sockets or such, a cycle will be created. We would not free this cycle
properly, because we treated any unit with non-empty refs as uncollectable. To
solve this issue, treats refs with UnitRef in u->refs_by_target similarly to
the refs in u->dependencies, and check if the "other" unit is known to be
needed. If it is not needed, do not treat the reference from it as preventing
the unit we are looking at from being freed.
"check" is unclear: what is true, what is false? Let's rename to "can_gc" and
revert the return value ("positive" values are easier to grok).
v2:
- rename from unit_can_gc to unit_may_gc
I think if we log the error as being _ignored_, we should also consider
the event as handled and clear it. This was the behaviour prior to
575b300b (PR #7968).
I don't think we particularly wanted to change behaviour and keep retrying.
Sometimes that's useful, other times you cause more problems by filling the
logs.
Plus a nearby typo fix.
Previously, we'd synchronize bus names immediately when we succeeded
connecting to the bus, potentially even before coldplugging the units.
This was problematic, as synchronizing bus names meant invoking the
per-unit name change handler function which might change the unit's
state — which will result in consistency when done before we coldplug
things.
With this change we instead enqueue a job for the event loop to resync
the names in a later loop iteration, i.e. at a point where we know
coldplugging has finished.
Let's also use the journal if it is currently reloading. In that state
it should also be able to process our requests. Moreover, we might
otherwise end up disconnecting/reconnecting from the journal without
really any need to hence, relax the check accordingly.
This removes the current bus_init() call, as it had multiple problems:
it munged handling of the three bus connections we care about (private,
"api" and system) into one, even though the conditions when which was
ready are very different. It also added redundant logging, as the
individual calls it called all logged on their own anyway.
The three calls bus_init_api(), bus_init_private() and bus_init_system()
are now made public. A new call manager_dbus_is_running() is added that
works much like manager_journal_is_running() and is a lot more careful
when checking whether dbus is around. Optionally it checks the unit's
deserialized_state rather than state, in order to accomodate for cases
where we cant to connect to the bus before deserializing the
"subscribed" list, before coldplugging the units.
manager_recheck_dbus() is added, that works a lot like
manager_recheck_journal() and is invoked in unit_notify(), i.e. when
units change state.
All in all this should make handling a bit more alike to journal
handling, and it also fixes one major bug: when running in user mode
we'll now connect to the system bus early on, without conditionalizing
this in anyway.
Let's simplify things a bit: we so far called both functions every
single time, let's just merge one into the other, so that we have fewer
functions to call.
After discussions with @htejun it appears it's OK now to enable memory
accounting by default for all units without affecting system performance
too badly. facebook has made good experiences with deploying memory
accounting across their infrastructure.
This hence turns MemoryAccounting= from opt-in to opt-out, similar to
how TasksAccounting= is already handled. The other accounting options
remain off, their performance impact is too big still.
Before this, each ExecRuntime object is owned by a unit. However,
it may be shared with other units which enable JoinsNamespaceOf=.
Thus, by the serialization/deserialization process, its sharing
information, more specifically, reference counter is lost, and
causes issue #7790.
This makes ExecRuntime objects be managed by manager, and changes
the serialization/deserialization process.
Fixes#7790.
log_open_console() did not switch from stderr to /dev/console, when
"always_reopen_console" was set. It was necessary to call
log_close_console() first.
By contrast, log_open() did switch between e.g. journald and kmsg according
to the value of "prohibit_ipc".
Let's fix log_open() to respect the values of all the log options, and we
can make log_close_*() private.
Also log_close_console() is changed. There was some precaution, avoiding
closing the console fd if we are not PID 1. I think commit 48a601fe made
a little mistake in leaving this in, and it only served to confuse
readers :).
Also I changed systemd-shutdown. Now we have log_set_prohibit_ipc(), let's
use it to clarify that systemd-shutdown is not expected to try and log via
journald (which it is about to kill). We avoided ever asking it to, but
it's more convenient for the reader if they don't have to think about that.
In that sense, it's similar to using assert() to validate a function's
arguments.
If we have to force the logging to close the journal fd, then we can open
any fallback log target. E.g. kmsg, if the target was the default
JOURNAL_OR_KMSG.
This is the behaviour I would expect from the documentation. I couldn't
find any justification in the code, for why we would want to start dropping
log messages instead of sending them to the fallback target.
This means we will match the behaviour of processes which we fork and which
set `open_when_needed`, and with generators - which use
log_set_prohibit_ipc(true) - which we fork+exec during a reload.
IMO this illustrates that the log_open/log_close interface is too clunky.
So with the behaviour settled, I will refactor the interface in the next
commit :).
Let's add a per-unit boolean that tells us whether our unit is currently
counted or not. This way it's unlikely we get out of sync again and
things are generally more robust.
This also allows us to remove the counting logic specific to service
units (which was in fact mostly a copy from the generic implementation),
in favour of fully generic code.
Replaces: #7824
Since the the whole function ultimately is just a fancy getter for the
show_status field, let's actually return it as last step literally
without an extra needless "if".
Previously, we'd maintain two hashmaps keyed by PIDs, pointing to Unit
interested in SIGCHLD events for them. This scheme allowed a specific
PID to be watched by exactly 0, 1 or 2 units.
With this rework this is replaced by a single hashmap which is primarily
keyed by the PID and points to a Unit interested in it. However, it
optionally also keyed by the negated PID, in which case it points to a
NULL terminated array of additional Unit objects also interested. This
scheme means arbitrary numbers of Units may now watch the same PID.
Runtime and memory behaviour should not be impact by this change, as for
the common case (i.e. each PID only watched by a single unit) behaviour
stays the same, but for the uncommon case (a PID watched by more than
one unit) we only pay with a single additional memory allocation for the
array.
Why this all? Primarily, because allowing exactly two units to watch a
specific PID is not sufficient for some niche cases, as processes can
belong to more than one unit these days:
1. sd_notify() with MAINPID= can be used to attach a process from a
different cgroup to multiple units.
2. Similar, the PIDFile= setting in unit files can be used for similar
setups,
3. By creating a scope unit a main process of a service may join a
different unit, too.
4. On cgroupsv1 we frequently end up watching all processes remaining in
a scope, and if a process opens lots of scopes one after the other it
might thus end up being watch by many of them.
This patch hence removes the 2-unit-per-PID limit. It also makes a
couple of other changes, some of them quite relevant:
- manager_get_unit_by_pid() (and the bus call wrapping it) when there's
ambiguity will prefer returning the Unit the process belongs to based on
cgroup membership, and only check the watch-pids hashmap if that
fails. This change in logic is probably more in line with what people
expect and makes things more stable as each process can belong to
exactly one cgroup only.
- Every SIGCHLD event is now dispatched to all units interested in its
PID. Previously, there was some magic conditionalization: the SIGCHLD
would only be dispatched to the unit if it was only interested in a
single PID only, or the PID belonged to the control or main PID or we
didn't dispatch a signle SIGCHLD to the unit in the current event loop
iteration yet. These rules were quite arbitrary and also redundant as
the the per-unit handlers would filter the PIDs anyway a second time.
With this change we'll hence relax the rules: all we do now is
dispatch every SIGCHLD event exactly once to each unit interested in
it, and it's up to the unit to then use or ignore this. We use a
generation counter in the unit to ensure that we only invoke the unit
handler once for each event, protecting us from confusion if a unit is
both associated with a specific PID through cgroup membership and
through the "watch_pids" logic. It also protects us from being
confused if the "watch_pids" hashmap is altered while we are
dispatching to it (which is a very likely case).
- sd_notify() message dispatching has been reworked to be very similar
to SIGCHLD handling now. A generation counter is used for dispatching
as well.
This also adds a new test that validates that "watch_pid" registration
and unregstration works correctly.
This fundamentally makes one change: we never process more than one
signal or more than one waitid() event per event loop. We'll never tight
loop around waitid() or around read() on our signalfd instead, but
always return to the main event loop after processing one event.
By doing this we put the event priorization handling into full power
again, as we'll always check for higher priority events before looking
at the next signal or waitid() again.
This introduces a new "defer" event source "sigchld_event". It's enabled
as soon as we see SIGCHLD, and disabled as soon as waitid() reported no
further children pending. It's running at a relatively high priority,
one step higher than signal handling itself, but lower than
/proc/self/mountinfo event handling, so that the latter always takes
precedence.
Since we want to process sd_notify() events at an even higher priority
than SIGCHLD (as before) it is moved one priority step up, too.
Fixes: #7932
Possibly fixes: #7966
Let's shorten manager_check_finished() a bit by splitting out checking
of basic.target and the two things we do when we reach it.
This should not change behaviour, except for one thing: we now check
basic.target's actual state for figuring out whether it is up, instead
of generically checking whether it has any job queued. This is arguably
more correct, and is what other code does too for similar purposes, for
example manager_state()
This happens to be almost the same moment as when we send READY=1 in the user
instance, but the logic is slightly different, since we log taint when
basic.target is reached in the system manager, but we send the notification
only in the user manager. So add a separate flag for this and propagate it
across reloads.
Fixes#7683.
Let's be more restrictive when validating PID files and MAINPID=
messages: don't accept PIDs that make no sense, and if the configuration
source is not trusted, don't accept out-of-cgroup PIDs. A configuratin
source is considered trusted when the PID file is owned by root, or the
message was received from root.
This should lock things down a bit, in case service authors write out
PID files from unprivileged code or use NotifyAccess=all with
unprivileged code. Note that doing so was always problematic, just now
it's a bit less problematic.
When we open the PID file we'll now use the CHASE_SAFE chase_symlinks()
logic, to ensure that we won't follow an unpriviled-owned symlink to a
privileged-owned file thinking this was a valid privileged PID file,
even though it really isn't.
Fixes: #6632
If we have to chose between truncated escape sequences and strings
exploded to 4 times the desried length by fully escaping, prefer the
latter.
It's for debug only, hence doesn't really matter much.
Let's rename it manager_sanitize_environment() which is a more precise
name. Moreover, sort the environment implicitly inside it, as all our
callers do that anyway afterwards and we can save some code this way.
Also, update the list of env vars to drop, i.e. the env vars we manage
ourselves and don't want user code to interfear with. Also sort this
list to make it easier to update later on.
This is useful so that callers know whether anything at all and how much
was flushed.
This patches through users of this functions to ensure that the return
values > 0 which may be returned now are not propagated in public APIs.
Also, users that ignore the return value are changed to do so explicitly
now.
This makes things a bit easier to read I think, and also makes sure we
always use the _unlikely_ wrapper around it, which so far we used
sometimes and other times we didn't. Let's clean that up.
We have a check and warning at compile time. The user cannot do anything about
this at runtime, and all other taints are about checks that happen at runtime
and are specific to that system (and at least potentially correctable).
(The logic in the compilation-time check was updated to treat "nogroup" as OK,
but not the runtime check. But I think it's better to remove the runtime check
for this altogether, so this becomes moot.)
Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f,
FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*,
instead of doing individual per-call decision whether to use normal
calls or _unlocked() calls.
This has various benefits:
1. It's easier to read and easier not to forget
2. It's more comprehensive, as fprintf() and friends are covered too
(as these functions have no _unlocked() counterpart)
3. Philosophically, it's a bit more correct, because it's more a
property of the file handle really whether we ever pass it on to another
thread, not of the operations we then apply to it.
This patch reworks all pieces of codes that so far used fxyz_unlocked()
calls to use __fsetlocking() instead. It also reworks all places that
use open_memstream(), i.e. use stdio FILE* for string manipulations.
Note that this in some way a revert of 4b61c87511.
Failure of systemd to respond on the bus interface was bisected to af6b0ecc
"core: make "taint" string logic a bit more generic and output it at boot".
Failure was presumably caused by trying to append strings to an
unintialized buffer, leading to writing outside the unterminated buffer
and hence undefined behaviour.
The tainting logic existed for a long time, but was hidden inside the
bus interfaces. Let's give it a small bit more coverage, by logging its
value early at boot during initialization.
This changes the unit search path logic to never drop the transient and
control directories from the unit search path. This is necessary as we
add new entries to both during runtime, due to the "systemctl
set-property" and transient unit logic.
Previously, the "transient" directory was created during early boot to
deal with this, but the "control" directories were not covered like
that. Creating the control directories early at boot is not possible
however, as /etc might be read-only then, and we do define a persistent
control directory. Hence, let's create these dirs on-demand when we need
them, and make sure the search path clean-up logic never drops them from
the search path even if they are initially missing.
(Also, always create these paths properly labelled)
In the user mode, not all special units exist.
So, we need to check whether the units exist or not before operate
something to the units.
Such the check was mistakenly dropped by e68537f0ba.
Fixes#7426.
Let's make use of unit_active_or_pending() where we can. Note that this
change changes beaviour in one specific case: when shutdown.target is
active we'll now also return that the system is in "stopping" state, not
only when we try to get into it. That makes sense as shutdown.target is
ordered before the actually shutdown units such as
"systemd-poweroff.service", and if the state is queried between reaching
those we should also report "stopping".
Let's make our finished checks a bit more readable. Checking the
timestamp is not entirely obvious, hence let's abstract that a bit by
adding a macro that shows what we are doing here, not how we doing it.
This is particularly useful if we want to change the definition of
"finished" later on, in particular, when we try to fix#7023.
It's like manager_dump(), but returns a string. This allows us to reduce
some duplicate code. Also, while we are at it, turn off stdio locking
while we write to the memory FILE *f.
This makes things quite a bit more systematic I think, as we can
systematically operate on all timestamps, for example for the purpose of
serialization/deserialization.
This rework doesn't necessarily make things shorter in the individual
lines, but it does reduce the line count a bit.
(This is useful particularly when we want to add additional timestamps,
for example to solve #7023)
Presets are useful to initialize uninitialized /etc, but that doesn't
apply to the initrd.
Also, let's rename etc_empty → first_boot. After all, the variable
doesn't actually reflect whether /etc is really empty, it just reflects
whether /etc/machine-id existed originally or not. Moreover, we later on
directly initialize manager_set_first_boot() from it, hence let's just
name it the same way all through the codepath, to make this all less
confusing.
See: #7100
And let's make use of it to implement two new unit settings with it:
1. LogLevelMax= is a new per-unit setting that may be used to configure
log priority filtering: set it to LogLevelMax=notice and only
messages of level "notice" and lower (i.e. more important) will be
processed, all others are dropped.
2. LogExtraFields= is a new per-unit setting for configuring per-unit
journal fields, that are implicitly included in every log record
generated by the unit's processes. It takes field/value pairs in the
form of FOO=BAR.
Also, related to this, one exisiting unit setting is ported to this new
facility:
3. The invocation ID is now pulled from /run/systemd/units/ instead of
cgroupfs xattrs. This substantially relaxes requirements of systemd
on the kernel version and the privileges it runs with (specifically,
cgroupfs xattrs are not available in containers, since they are
stored in kernel memory, and hence are unsafe to permit to lesser
privileged code).
/run/systemd/units/ is a new directory, which contains a number of files
and symlinks encoding the above information. PID 1 creates and manages
these files, and journald reads them from there.
Note that this is supposed to be a direct path between PID 1 and the
journal only, due to the special runtime environment the journal runs
in. Normally, today we shouldn't introduce new interfaces that (mis-)use
a file system as IPC framework, and instead just an IPC system, but this
is very hard to do between the journal and PID 1, as long as the IPC
system is a subject PID 1 manages, and itself a client to the journal.
This patch cleans up a couple of types used in journal code:
specifically we switch to size_t for a couple of memory-sizing values,
as size_t is the right choice for everything that is memory.
Fixes: #4089Fixes: #3041Fixes: #4441
We would continue, but still return an error at the end. This isn't useful
because we'd still error-out in main().
Also, add a missing error message when we fail to mkdir.
This replaces the dependencies Set* objects by Hashmap* objects, where
the key is the depending Unit, and the value is a bitmask encoding why
the specific dependency was created.
The bitmask contains a number of different, defined bits, that indicate
why dependencies exist, for example whether they are created due to
explicitly configured deps in files, by udev rules or implicitly.
Note that memory usage is not increased by this change, even though we
store more information, as we manage to encode the bit mask inside the
value pointer each Hashmap entry contains.
Why this all? When we know how a dependency came to be, we can update
dependencies correctly when a configuration source changes but others
are left unaltered. Specifically:
1. We can fix UDEV_WANTS dependency generation: so far we kept adding
dependencies configured that way, but if a device lost such a
dependency we couldn't them again as there was no scheme for removing
of dependencies in place.
2. We can implement "pin-pointed" reload of unit files. If we know what
dependencies were created as result of configuration in a unit file,
then we know what to flush out when we want to reload it.
3. It's useful for debugging: "systemd-analyze dump" now shows
this information, helping substantially with understanding how
systemd's dependency tree came to be the way it came to be.
manager_connect_bus() is called *before* manager_coldplug(). As a last
thing in service_coldplug() we set service state to
s->deserialized_state, and thus before we do that all services are
inactive and try_connect always evaluates to false. To fix that we must
look at deserialized state instead of current unit state.
Fixes#7146
When a user logs in, systemd-pam will wait for the user manager instance to
report readiness. We don't need to wait for all the jobs to finish, it
is enough if the basic startup is done and the user manager is responsive.
systemd --user will now send out a READY=1 notification when either of two
conditions becomes true:
- basic.target/start job is gone,
- the initial transaction is done.
Also fixes#2863.
The advantage is that is the name is mispellt, cpp will warn us.
$ git grep -Ee "conf.set\('(HAVE|ENABLE)_" -l|xargs sed -r -i "s/conf.set\('(HAVE|ENABLE)_/conf.set10('\1_/"
$ git grep -Ee '#ifn?def (HAVE|ENABLE)' -l|xargs sed -r -i 's/#ifdef (HAVE|ENABLE)/#if \1/; s/#ifndef (HAVE|ENABLE)/#if ! \1/;'
$ git grep -Ee 'if.*defined\(HAVE' -l|xargs sed -i -r 's/defined\((HAVE_[A-Z0-9_]*)\)/\1/g'
$ git grep -Ee 'if.*defined\(ENABLE' -l|xargs sed -i -r 's/defined\((ENABLE_[A-Z0-9_]*)\)/\1/g'
+ manual changes to meson.build
squash! build-sys: use #if Y instead of #ifdef Y everywhere
v2:
- fix incorrect setting of HAVE_LIBIDN2
In most cases we followed the rule that the special _INVALID and _MAX
values we use in our enums use the full type name as prefix (in contrast
to regular values that we often make shorter), do so for
ExecDirectoryType as well.
No functional changes, just a little bit of renaming to make this code
more like the rest.
This makes sure that if we learn via inotify or another event source
that a cgroup is empty, and we checked that this is indeed the case (as
we might get spurious notifications through inotify, as the inotify
logic through the "cgroups.event" is pretty unspecific and might be
trigger for a variety of reasons), then we'll enqueue a defer event for
it, at a priority lower than SIGCHLD handling, so that we know for sure
that if there's waitid() data for a process we used it before
considering the cgroup empty notification.
Fixes: #6608
We are about to add second cgroup-related queue, called
"cgroup_empty_queue", hence let's rename "cgroup_queue" to
"cgroup_realize_queue" (as that is its purpose) to minimize confusion
about the two queues.
Just a rename, no functional changes.
This way we can safely run manager objects from tests and good timeouts
apply. Without this all timeouts are set 0, which means they fire
instantly, when run from tests which do not explicitly configure them
(the way main.c does).
Now generators are only run in systemd --test mode, where this makes
most sense (how are you going to test what would happen otherwise?).
Fixes#6842.
v2:
- rename test_run to test_run_flags
When running through systemd-analyze verify or with --test, we would
not run generators (environment or unit). But at the end, we would nuke
the generator dirs anyway.
Simplify things by actually running generators of both types, but redirecting
their output to a temporary directory. This has the advantage that we test more
code, and the verification is more complete.
Since now we are not touching the real generator directories, we also don't
delete them, which fixes#5609.
When running in systemd-analyze verify, first_boot was initialized to -1
and never changed, so we'd try to run unit_file_preset_all(). Change the
check to > 0 which is more correct. Also, add a separate test for !test_run,
since we wouldn't want to run presets even if we were in first boot
(or /etc was empty for whatever other reason).
If dbus is already down during shutdown, we can't propagate the cgroup
release message anymore, but that's expected and nothing to warn about.
Hence let's downgrade the message from LOG_WARN to LOG_DEBUG.
Fixes: #6777
Trying to connect otherwise is pointless, because if socket isn't around
we won't connect. However, when dbus.socket is present we attempt to
connect. That attempt can't succeed because we are then supposed
to activate dbus.service as a response to connection from
us. This results in deadlock.
Fixes#6303
The irreversible job mode is required to ensure that shutdown is not
interrupted by the activation of a unit with a conflict.
We already used the correct job mode for `ctrl-alt-del.target`. But not
for `exit.target` (SIGINT of user manager). The SIGRT shutdown signals
also needed fixing.
Also change SIGRTMIN+0 to isolate default.target, instead of starting
it. The previous behaviour was documented. However there was no reason
given for it, nor can we provide one. The problem that isolate is too
aggressive anywhere outside of emergency.target (#2607) is orthogonal.
This feature is "accessible by different means and only really a safety
net"; it is confusing for it to differ from `systemctl default` without
explanation.
`AllowIsolate=yes` is retained on poweroff.target etc. for backwards
compatibility.
`sigpwr.target` is also an obvious candidate for linking to a shutdown
target. Unforunately it is also a possible hook for implementing some
logic like system V init did, reading `/etc/powerstatus`. If we switched
to starting `sigpwr.target` with REPLACE_IRREVERSIBLY, attempts to run
`systemctl shutdown` from it would fail, if they had not thought to set
`DefaultDependencies=no`. We had provided no examples for `sigpwr`, and
the whole idea is cruft to keep legacy people happy. For the moment, I
leave `sigpwr` alone, with no risk of disrupting anyone's
previously-working, half-working, or untested setup.
Fixes#6484. See also #6471
The comment here was misleading: the job can fail to enqueue for reasons
other than the target not existing.
The fallback caused an error to be logged, and dates back to when the
"user" directory was named "session". units/session/exit.target was added
later the same year.
This is consistent with the documentation (man systemd), and the handling
of similar signals. It's also consistent with `systemctl exit`, which is
what most people would expect.
Let's decouple the Manager object from the execution logic a bit more
here too, and simply pass along the fact whether we should
unconditionally chown the runtime/... directories via the ExecFlags
field too.
Let's try to decouple the execution engine a bit from the Unit/Manager
concept, and hence pass one more flag as part of the ExecParameters flags
field.
This commit moves the first-boot system preset-settings evaluation out
of main and into the manager startup logic itself. Notably, it reverses
the order between generators and presets evaluation, so that any changes
performed by first-boot generators are taken into the account by presets
logic.
After this change, units created by a generator can be enabled as part
of a preset.
We just abort startup, without printing any error. Make sure we always
print something, and when we cannot deserialize some unit, just ignore it and
continue.
Fixup for 4bc5d27b94. Without this, we would hang
in daemon-reexec after upgrade.
As a follow-up for db3f45e2d2 let's do the
same for all other cases where we create a FILE* with local scope and
know that no other threads hence can have access to it.
For most cases this shouldn't change much really, but this should speed
dbus introspection and calender time formatting up a bit.
This introduces {State,Cache,Log,Configuration}Directory= those are
similar to RuntimeDirectory=. They create the directories under
/var/lib, /var/cache/, /var/log, or /etc, respectively, with the mode
specified in {State,Cache,Log,Configuration}DirectoryMode=.
This also fixes#6391.
Apart from bugs (as in #6152), this can happen if we ever make
our requirements for environment entries more stringent. As with
the rest of deserialization, we should just warn and continue.
We use our cgroup APIs in various contexts, including from our libraries
sd-login, sd-bus. As we don#t control those environments we can't rely
that the unified cgroup setup logic succeeds, and hence really shouldn't
assert on it.
This more or less reverts 415fc41cea.
Environment file generators are a lot like unit file generators, but not
exactly:
1. environment file generators are run for each manager instance, and their
output is (or at least can be) individualized.
The generators themselves are system-wide, the same for all users.
2. environment file generators are run sequentially, in priority order.
Thus, the lifetime of those files is tied to lifecycle of the manager
instance. Because generators are run sequentially, later generators can use or
modify the output of earlier generators.
Each generator is run with no arguments, and the whole state is stored in the
environment variables. The generator can echo a set of variable assignments to
standard output:
VAR_A=something
VAR_B=something else
This output is parsed, and the next and subsequent generators run with those
updated variables in the environment. After the last generator is done, the
environment that the manager itself exports is updated.
Each generator must return 0, otherwise the output is ignored.
The generators in */user-env-generator are for the user session managers,
including root, and the ones in */system-env-generator are for pid1.
This protocol is generally useful, we might just as well reuse it for the
env. generators.
The implementation is changed a bit: instead of making a new strv and freeing
the old one, just mutate the original. This is much faster with larger arrays,
while in fact atomicity is preserved, since we only either insert the new
entry or not, without being in inconsistent state.
v2:
- fix confusion with return value
The output of processes can be gathered, and passed back to the callee.
(This commit just implements the basic functionality and tests.)
After the preparation in previous commits, the change in functionality is
relatively simple. For coding convenience, alarm is prepared *before* any
children are executed, and not before. This shouldn't matter usually, since
just forking of the children should be pretty quick. One could also argue that
this is more correct, because we will also catch the case when (for whatever
reason), forking itself is slow.
Three callback functions and three levels of serialization are used:
- from individual generator processes to the generator forker
- from the forker back to the main process
- deserialization in the main process
v2:
- replace an structure with an indexed array of callbacks
There is a slight change in behaviour: the user manager for root will create a
temporary file in /run/systemd, not /tmp. I don't think this matters, but
simplifies implementation.
cg_[all_]unified() test whether a specific controller or all controllers are on
the unified hierarchy. While what's being asked is a simple binary question,
the callers must assume that the functions may fail any time, which
unnecessarily complicates their usages. This complication is unnecessary.
Internally, the test result is cached anyway and there are only a few places
where the test actually needs to be performed.
This patch simplifies cg_[all_]unified().
* cg_[all_]unified() are updated to return bool. If the result can't be
decided, assertion failure is triggered. Error handlings from their callers
are dropped.
* cg_unified_flush() is updated to calculate the new result synchrnously and
return whether it succeeded or not. Places which need to flush the test
result are updated to test for failure. This ensures that all the following
cg_[all_]unified() tests succeed.
* Places which expected possible cg_[all_]unified() failures are updated to
call and test cg_unified_flush() before calling cg_[all_]unified(). This
includes functions used while setting up mounts during boot and
manager_setup_cgroup().
Embedding sd_id128_t's in constant strings was rather cumbersome. We had
SD_ID128_CONST_STR which returned a const char[], but it had two problems:
- it wasn't possible to statically concatanate this array with a normal string
- gcc wasn't really able to optimize this, and generated code to perform the
"conversion" at runtime.
Because of this, even our own code in coredumpctl wasn't using
SD_ID128_CONST_STR.
Add a new macro to generate a constant string: SD_ID128_MAKE_STR.
It is not as elegant as SD_ID128_CONST_STR, because it requires a repetition
of the numbers, but in practice it is more convenient to use, and allows gcc
to generate smarter code:
$ size .libs/systemd{,-logind,-journald}{.old,}
text data bss dec hex filename
1265204 149564 4808 1419576 15a938 .libs/systemd.old
1260268 149564 4808 1414640 1595f0 .libs/systemd
246805 13852 209 260866 3fb02 .libs/systemd-logind.old
240973 13852 209 255034 3e43a .libs/systemd-logind
146839 4984 34 151857 25131 .libs/systemd-journald.old
146391 4984 34 151409 24f71 .libs/systemd-journald
It is also much easier to check if a certain binary uses a certain MESSAGE_ID:
$ strings .libs/systemd.old|grep MESSAGE_ID
MESSAGE_ID=%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x
MESSAGE_ID=%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x
MESSAGE_ID=%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x
MESSAGE_ID=%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x
$ strings .libs/systemd|grep MESSAGE_ID
MESSAGE_ID=c7a787079b354eaaa9e77b371893cd27
MESSAGE_ID=b07a249cd024414a82dd00cd181378ff
MESSAGE_ID=641257651c1b4ec9a8624d7a40a9e1e7
MESSAGE_ID=de5b426a63be47a7b6ac3eaac82e2f6f
MESSAGE_ID=d34d037fff1847e6ae669a370e694725
MESSAGE_ID=7d4958e842da4a758f6c1cdc7b36dcc5
MESSAGE_ID=1dee0369c7fc4736b7099b38ecb46ee7
MESSAGE_ID=39f53479d3a045ac8e11786248231fbf
MESSAGE_ID=be02cf6855d2428ba40df7e9d022f03d
MESSAGE_ID=7b05ebc668384222baa8881179cfda54
MESSAGE_ID=9d1aaa27d60140bd96365438aad20286
At -O3, this was printed a hundred times for various callers of
manager_add_job_by_name(). AFAICT, there is no error and `unit` is always
intialized. Nevertheless, add explicit initialization to silence the noise.
src/core/manager.c: In function 'manager_start_target':
src/core/manager.c:1413:16: warning: 'unit' may be used uninitialized in this function [-Wmaybe-uninitialized]
return manager_add_job(m, type, unit, mode, e, ret);
^
src/core/manager.c:1401:15: note: 'unit' was declared here
Unit *unit;
^
We were inconsistent, manager_load_unit_prepare() would crash if _ret was ever NULL.
But none of the callers use NULL. So simplify things and require it to be non-NULL.
If we can, use a memfd for serializing state during a daemon reload or
reexec. Fall back to a file in /run/systemd or /tmp only if memfds are
not available.
See: #5016
That message is emitted by every systemd instance on every resume:
Dec 06 08:03:38 laptop systemd[1]: Time has been changed
Dec 06 08:03:38 laptop systemd[823]: Time has been changed
Dec 06 08:03:38 laptop systemd[916]: Time has been changed
Dec 07 08:00:32 laptop systemd[1]: Time has been changed
Dec 07 08:00:32 laptop systemd[823]: Time has been changed
Dec 07 08:00:32 laptop systemd[916]: Time has been changed
-- Reboot --
Dec 07 08:02:46 laptop systemd[836]: Time has been changed
Dec 07 08:02:46 laptop systemd[1]: Time has been changed
Dec 07 08:02:46 laptop systemd[926]: Time has been changed
Dec 07 19:48:12 laptop systemd[1]: Time has been changed
Dec 07 19:48:12 laptop systemd[836]: Time has been changed
Dec 07 19:48:12 laptop systemd[926]: Time has been changed
...
Fixes#4896.
This add a new message id for the end of user instance startup.
User manager startup is a different beast then the system startup.
Their descriptions are completely different too. Let's just separate
them.
Partially fixes#3351.
Also remove "successful" from the description, since we don't know if
the startup was successful or not.
It's rather hard to parse the confirmation messages (enabled with
systemd.confirm_spawn=true) amongst the status messages and the kernel
ones (if enabled).
This patch gives the possibility to the user to redirect the confirmation
message to a different virtual console, either by giving its name or its path,
so those messages are separated from the other ones and easier to read.
When booting with systemd.confirm_spawn=true, the eye of cylon
animation kicks in pretty quickly so user doesn't have any chance to
answer the questions which services to start before the confirmation
message is screwed by the cylon.
This basically breaks the confirm_spawn functionality completely.
This patch prevents the cylon animation to kick in when
confirmation_spawn=yes.
Fixes: #2194
In contrast to all other unit types device units when queued just track
external state, they cannot effect state changes on their own. Hence unless a
client or other job waits for them there's no reason to keep them in the job
queue. This adds a concept of GC'ing jobs of this type as soon as no client or
other job waits for them anymore.
To ensure this works correctly we need to track which clients actually
reference a job (i.e. which ones enqueued it). Unfortunately that's pretty
nasty to do for direct connections, as sd_bus_track doesn't work for
them. For now, work around this, by simply remembering in a boolean that a job
was requested by a direct connection, and reset it when we notice the direct
connection is gone. This means the GC logic works fine, except that jobs are
not immediately removed when direct connections disconnect.
In the longer term, a rework of the bus logic should fix this properly. For now
this should be good enough, as GC works for fine all cases except this one, and
thus is a clear improvement over the previous behaviour.
Fixes: #1921
I think it's an antipattern to have to count the number of bytes in
the prefix by hand. We should do this automatically to avoid wasting
programmer time, and possible errors. I didn't any offsets that were
wrong, so this change is mostly to make future development easier.
Since we ignore the result anyway, downgrade errors to warning.
log_oom() will still emit an error, but that's mostly theoretical, so it
is not worth complicating the code to avoid the small inconsistency
This adds a new invocation ID concept to the service manager. The invocation ID
identifies each runtime cycle of a unit uniquely. A new randomized 128bit ID is
generated each time a unit moves from and inactive to an activating or active
state.
The primary usecase for this concept is to connect the runtime data PID 1
maintains about a service with the offline data the journal stores about it.
Previously we'd use the unit name plus start/stop times, which however is
highly racy since the journal will generally process log data after the service
already ended.
The "invocation ID" kinda matches the "boot ID" concept of the Linux kernel,
except that it applies to an individual unit instead of the whole system.
The invocation ID is passed to the activated processes as environment variable.
It is additionally stored as extended attribute on the cgroup of the unit. The
latter is used by journald to automatically retrieve it for each log logged
message and attach it to the log entry. The environment variable is very easily
accessible, even for unprivileged services. OTOH the extended attribute is only
accessible to privileged processes (this is because cgroupfs only supports the
"trusted." xattr namespace, not "user."). The environment variable may be
altered by services, the extended attribute may not be, hence is the better
choice for the journal.
Note that reading the invocation ID off the extended attribute from journald is
racy, similar to the way reading the unit name for a logging process is.
This patch adds APIs to read the invocation ID to sd-id128:
sd_id128_get_invocation() may be used in a similar fashion to
sd_id128_get_boot().
PID1's own logging is updated to always include the invocation ID when it logs
information about a unit.
A new bus call GetUnitByInvocationID() is added that allows retrieving a bus
path to a unit by its invocation ID. The bus path is built using the invocation
ID, thus providing a path for referring to a unit that is valid only for the
current runtime cycleof it.
Outlook for the future: should the kernel eventually allow passing of cgroup
information along AF_UNIX/SOCK_DGRAM messages via a unique cgroup id, then we
can alter the invocation ID to be generated as hash from that rather than
entirely randomly. This way we can derive the invocation race-freely from the
messages.
Let's not accept datagrams with embedded NUL bytes. Previously we'd simply
ignore everything after the first NUL byte. But given that sending us that is
pretty ugly let's instead complain and refuse.
With this change we'll only accept messages that have exactly zero or one NUL
bytes at the very end of the datagram.
Let's make the kernel let us know the full, original datagram size of the
incoming message. If it's larger than the buffer space provided by us, drop the
whole message with a warning.
Before this change the kernel would truncate the message for us to the buffer
space provided, and we'd not complain about this, and simply process the
incomplete message as far as it made sense.
If the kernel doesn't permit us to dequeue/process an incoming notification
datagram message it's still better to stop processing the notification messages
altogether than to enter a busy loop where we keep getting notified but can't
do a thing about it.
With this change, manager_dispatch_notify_fd() behaviour is changed like this:
- if an error indicating a spurious wake-up is seen on recvmsg(), ignore it
(EAGAIN/EINTR)
- if any other error is seen on recvmsg() propagate it, thus disabling
processing of further wakeups
- if any error is seen on later code in the function, warn about it but do not
propagate it, as in this cas we're not going to busy loop as the offending
message is already dequeued.
For some certification, it should not be possible to reboot the machine through ctrl-alt-delete. Currently we suggest our customers to mask the ctrl-alt-delete target, but that is obviously not enough.
Patching the keymaps to disable that is really not a way to go for them, because the settings need to be easily checked by some SCAP tools.
This prevented systemd-analyze from unprivileged operation on older systemd
installations, which should be possible.
Also, we shouldn't touch the file system in test mode even if we can.
"closing all" might suggest that _all_ fds received with the notification message
will be closed. Reword the message to clarify that only the "unused" ones will be
closed.
This undoes 531ac2b234. I acked that patch without looking at the code
carefully enough. There are two problems:
- we want to process the fds anyway
- in principle empty notification messages are valid, and we should
process them as usual, including logging using log_unit_debug().
If manager_dispatch_notify_fd() fails and returns an error then the handling of
service notifications will be disabled entirely leading to a compromised system.
For example pid1 won't be able to receive the WATCHDOG messages anymore and
will kill all services supposed to send such messages.
This adds two (privileged) bus calls Ref() and Unref() to the Unit interface.
The two calls may be used by clients to pin a unit into memory, so that various
runtime properties aren't flushed out by the automatic GC. This is necessary
to permit clients to race-freely acquire runtime results (such as process exit
status/code or accumulated CPU time) on successful service termination.
Ref() and Unref() are fully recursive, hence act like the usual reference
counting concept in C. Taking a reference is a privileged operation, as this
allows pinning units into memory which consumes resources.
Transient units may also gain a reference at the time of creation, via the new
AddRef property (that is only defined for transient units at the time of
creation).
This adds the boolean RemoveIPC= setting to service, socket, mount and swap
units (i.e. all unit types that may invoke processes). if turned on, and the
unit's user/group is not root, all IPC objects of the user/group are removed
when the service is shut down. The life-cycle of the IPC objects is hence bound
to the unit life-cycle.
This is particularly relevant for units with dynamic users, as it is essential
that no objects owned by the dynamic users survive the service exiting. In
fact, this patch adds code to imply RemoveIPC= if DynamicUser= is set.
In order to communicate the UID/GID of an executed process back to PID 1 this
adds a new "user lookup" socket pair, that is inherited into the forked
processes, and closed before the exec(). This is needed since we cannot do NSS
from PID 1 due to deadlock risks, However need to know the used UID/GID in
order to clean up IPC owned by it if the unit shuts down.
Currently, systemd uses either the legacy hierarchies or the unified hierarchy.
When the legacy hierarchies are used, systemd uses a named legacy hierarchy
mounted on /sys/fs/cgroup/systemd without any kernel controllers for process
management. Due to the shortcomings in the legacy hierarchy, this involves a
lot of workarounds and complexities.
Because the unified hierarchy can be mounted and used in parallel to legacy
hierarchies, there's no reason for systemd to use a legacy hierarchy for
management even if the kernel resource controllers need to be mounted on legacy
hierarchies. It can simply mount the unified hierarchy under
/sys/fs/cgroup/systemd and use it without affecting other legacy hierarchies.
This disables a significant amount of fragile workaround logics and would allow
using features which depend on the unified hierarchy membership such bpf cgroup
v2 membership test. In time, this would also allow deleting the said
complexities.
This patch updates systemd so that it prefers the unified hierarchy for the
systemd cgroup controller hierarchy when legacy hierarchies are used for kernel
resource controllers.
* cg_unified(@controller) is introduced which tests whether the specific
controller in on unified hierarchy and used to choose the unified hierarchy
code path for process and service management when available. Kernel
controller specific operations remain gated by cg_all_unified().
* "systemd.legacy_systemd_cgroup_controller" kernel argument can be used to
force the use of legacy hierarchy for systemd cgroup controller.
* nspawn: By default nspawn uses the same hierarchies as the host. If
UNIFIED_CGROUP_HIERARCHY is set to 1, unified hierarchy is used for all. If
0, legacy for all.
* nspawn: arg_unified_cgroup_hierarchy is made an enum and now encodes one of
three options - legacy, only systemd controller on unified, and unified. The
value is passed into mount setup functions and controls cgroup configuration.
* nspawn: Interpretation of SYSTEMD_CGROUP_CONTROLLER to the actual mount
option is moved to mount_legacy_cgroup_hierarchy() so that it can take an
appropriate action depending on the configuration of the host.
v2: - CGroupUnified enum replaces open coded integer values to indicate the
cgroup operation mode.
- Various style updates.
v3: Fixed a bug in detect_unified_cgroup_hierarchy() introduced during v2.
v4: Restored legacy container on unified host support and fixed another bug in
detect_unified_cgroup_hierarchy().
A following patch will update cgroup handling so that the systemd controller
(/sys/fs/cgroup/systemd) can use the unified hierarchy even if the kernel
resource controllers are on the legacy hierarchies. This would require
distinguishing whether all controllers are on cgroup v2 or only the systemd
controller is. In preparation, this patch renames cg_unified() to
cg_all_unified().
This patch doesn't cause any functional changes.
This adds a new boolean setting DynamicUser= to service files. If set, a new
user will be allocated dynamically when the unit is started, and released when
it is stopped. The user ID is allocated from the range 61184..65519. The user
will not be added to /etc/passwd (but an NSS module to be added later should
make it show up in getent passwd).
For now, care should be taken that the service writes no files to disk, since
this might result in files owned by UIDs that might get assigned dynamically to
a different service later on. Later patches will tighten sandboxing in order to
ensure that this cannot happen, except for a few selected directories.
A simple way to test this is:
systemd-run -p DynamicUser=1 /bin/sleep 99999
As it turns out 512 is max number of tasks per service is hit by too many
applications, hence let's bump it a bit, and make it relative to the system's
maximum number of PIDs. With this change the new default is 15%. At the
kernel's default pids_max value of 32768 this translates to 4915. At machined's
default TasksMax= setting of 16384 this translates to 2457.
Why 15%? Because it sounds like a round number and is close enough to 4096
which I was going for, i.e. an eight-fold increase over the old 512
Summary:
| on the host | in a container
old default | 512 | 512
new default | 4915 | 2457
During stop when service has one "regular" pid one main pid and one
control pid and the sighld for the regular one is processed first the
unit_tidy_watch_pids will skip the main and control pid and does not
remove them from u->pids(). But then we skip the sigchld event because we
already did one in the iteration and there are two pids in u->pids.
v2: Use general unit_main_pid() and unit_control_pid() instead of
reaching directly to service structure.
By default, each iteration of manager_dispatch_sigchld() results in a unit level
sigchld event being invoked. For scope units, this results in a scope_sigchld_event()
which can seemingly stall for workloads that have a large number of PIDs within the
scope. The stall exhibits itself as a SIG_0 being initiated for each u->pids entry
as a result of pid_is_unwaited().
v2:
This patch resolves this condition by only paying to cost of a sigchld in the underlying
scope unit once per sigchld iteration. A new "sigchldgen" member resides within the
Unit struct. The Manager is incremented via the sd event loop, accessed via
sd_event_get_iteration, and the Unit member is set to the same value as the manager each
time that a sigchld event is invoked. If the Manager iteration value and Unit member
match, the sigchld event is not invoked for that iteration.
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.
When unit is marked as UNSURE, we are trying to find if it state was
changed over and over again. So lets not go through the UNSURE states
again. Also when we find a GOOD unit lets propagate the GOOD state to
all units that this unit reference.
This is a problem on machines with a lot of initscripts with different
starting priority, since those units will reference each other and the
original algorithm might get to n! complexity.
Thanks HATAYAMA Daisuke for the expand_good_state code.
We currently generate log message about unit being started even when
unit was started already and job didn't do anything. This is because job
was requested explicitly and hence became anchor job of the transaction
thus we could not eliminate it. That is fine but, let's not pollute
journal with useless log messages.
$ systemctl start systemd-resolved
$ systemctl start systemd-resolved
$ systemctl start systemd-resolved
Current state:
$ journalctl -u systemd-resolved | grep Started
May 05 15:31:42 rawhide systemd[1]: Started Network Name Resolution.
May 05 15:31:59 rawhide systemd[1]: Started Network Name Resolution.
May 05 15:32:01 rawhide systemd[1]: Started Network Name Resolution.
After patch applied:
$ journalctl -u systemd-resolved | grep Started
May 05 16:42:12 rawhide systemd[1]: Started Network Name Resolution.
Fixes#1723
The macro determines the right length of a AF_UNIX "struct sockaddr_un" to pass to
connect() or bind(). It automatically figures out if the socket refers to an
abstract namespace socket, or a socket in the file system, and properly handles
the full length of the path field.
This macro is not only safer, but also simpler to use, than the usual
offsetof() + strlen() logic.
dbus-daemon currently uses a backlog of 30 on its D-bus system bus socket. On
overloaded systems this means that only 30 connections may be queued without
dbus-daemon processing them before further connection attempts fail. Our
cgroups-agent binary so far used D-Bus for its messaging, and hitting this
limit hence may result in us losing cgroup empty messages.
This patch adds a seperate cgroup agent socket of type AF_UNIX/SOCK_DGRAM.
Since sockets of these types need no connection set up, no listen() backlog
applies. Our cgroup-agent binary will hence simply block as long as it can't
enqueue its datagram message, so that we won't lose cgroup empty messages as
likely anymore.
This also rearranges the ordering of the processing of SIGCHLD signals, service
notification messages (sd_notify()...) and the two types of cgroup
notifications (inotify for the unified hierarchy support, and agent for the
classic hierarchy support). We now always process events for these in the
following order:
1. service notification messages (SD_EVENT_PRIORITY_NORMAL-7)
2. SIGCHLD signals (SD_EVENT_PRIORITY_NORMAL-6)
3. cgroup inotify and cgroup agent (SD_EVENT_PRIORITY_NORMAL-5)
This is because when receiving SIGCHLD we invalidate PID information, which we
need to process the service notification messages which are bound to PIDs.
Hence the order between the first two items. And we want to process SIGCHLD
metadata to detect whether a service is gone, before using cgroup
notifications, to decide when a service is gone, since the former carries more
useful metadata.
Related to this:
https://bugs.freedesktop.org/show_bug.cgi?id=95264https://github.com/systemd/systemd/issues/1961
This moves the O_TMPFILE handling from the coredumping code into common library
code, and generalizes it as open_tmpfile_linkable() + link_tmpfile(). The
existing open_tmpfile() function (which creates an unlinked temporary file that
cannot be linked into the fs) is renamed to open_tmpfile_unlinkable(), to make
the distinction clear. Thus, code may now choose between:
a) open_tmpfile_linkable() + link_tmpfile()
b) open_tmpfile_unlinkable()
Depending on whether they want a file that may be linked back into the fs later
on or not.
In a later commit we should probably convert fopen_temporary() to make use of
open_tmpfile_linkable().
Followup for: #3065
The SysV compat code checks whether there's a native unit file before looking
for a SysV init script. Since the newest rework generated units will show up in
the unit path, and hence the checks ended up assuming that there always was a
native unit file for each init script: the generated one.
With this change the generated unit file directory is suppressed from the
search path when this check is done, to avoid the confusion.
This is too confusing, as this funciton returns the paths to the generator
binaries, while usually when we refer to the just the "generator path" we mean
the generated unit files. Let's clean this up.
Previously, transient units were created below the normal runtime directory
/run/systemd/system. With this change they are created in a special transient
directory /run/systemd/transient, which only contains data for transient units.
This clarifies the life-cycle of transient units, and makes clear they are
distinct from user-provided runtime units. In particular, users may now
extend transient units via /run/systemd/system, without systemd interfering
with the life-cycle of these files.
This change also adds code so that when a transient unit exits only the
drop-ins in this new directory are removed, but nothing else.
Fixes: #2139
This replaces the old function call manager_is_reloading_or_reexecuting() which
was used only at very few places. Use the new macro wherever we check whether
we are reloading. This should hopefully make things a bit more readable, given
the nature of Manager:n_reloading being a counter.
Previously, we had two enums ManagerRunningAs and UnitFileScope, that were
mostly identical and converted from one to the other all the time. The latter
had one more value UNIT_FILE_GLOBAL however.
Let's simplify things, and remove ManagerRunningAs and replace it by
UnitFileScope everywhere, thus making the translation unnecessary. Introduce
two new macros MANAGER_IS_SYSTEM() and MANAGER_IS_USER() to simplify checking
if we are running in one or the user context.
A long time ago – when generators where first introduced – the directories for
them were randomly created via mkdtemp(). This was changed later so that they
use fixed name directories now. Let's make use of this, and add the genrator
dirs to the LookupPaths structure and into the unit file search path maintained
in it. This has the benefit that the generator dirs are now normal part of the
search path for all tools, and thus are shown in "systemctl list-unit-files"
too.
After receiving SIGCHLD, one of the ways manager_dispatch_sigchld() maps the
now zombie $PID to its unit is through manager_get_unit_by_pid_cgroup() which
reads /proc/$PID/cgroup and looks up the unit associated with the cgroup path.
On non-unified cgroup hierarchies, a process is immediately migrated to the
root cgroup on death and the cgroup lookup would always have returned the unit
associated with it, making it rather pointless but safe. On unified hierarchy,
a zombie remains associated with the cgroup that it was associated with at the
time of death and thus manager_get_unit_by_pid_cgroup() will look up the unit
properly.
However, by the time manager_dispatch_sigchld() is running, the original cgroup
may have become empty and it and its associated unit might already have been
removed. If the cgroup path doesn't yield a match, manager_dispatch_sigchld()
keeps pruning the leaf component. This means that the function may return a
slice unit for a pid and as a slice doesn't have ->sigchld_event() handler,
calling invoke_sigchld_event() on it causes a segfault.
This patch updates invoke_sigchld_event() so that it skips calling if the
handler is not set.
Throughout the tree there's spurious use of spaces separating ++ and --
operators from their respective operands. Make ++ and -- operator
consistent with the majority of existing uses; discard the spaces.
Support for net_cls.class_id through the NetClass= configuration directive
has been added in v227 in preparation for a per-unit packet filter mechanism.
However, it turns out the kernel people have decided to deprecate the net_cls
and net_prio controllers in v2. Tejun provides a comprehensive justification
for this in his commit, which has landed during the merge window for kernel
v4.5:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
As we're aiming for full support for the v2 cgroup hierarchy, we can no
longer support this feature. Userspace tool such as nftables are moving over
to setting rules that are specific to the full cgroup path of a task, which
obsoletes these controllers anyway.
This commit removes support for tweaking details in the net_cls controller,
but keeps the NetClass= directive around for legacy compatibility reasons.
When we determine the current system state we check whether units like emergency.target are running or a job that
results in them being run is queued. However, this is not the case for JOB_TRY_RESTART, since that's a NOP if the unit
has not been running before. Hence, don't bother with checking for that job type.
==1== HEAP SUMMARY:
==1== in use at exit: 61,728 bytes in 22 blocks
==1== total heap usage: 258,122 allocs, 258,100 frees, 78,219,628
bytes allocated
==1==
==1== 16 bytes in 1 blocks are definitely lost in loss record 1 of 6
==1== at 0x4C2BBCF: malloc (vg_replace_malloc.c:299)
==1== by 0x1E350E: memdup (alloc-util.c:34)
==1== by 0x135AFB: memdup_multiply (alloc-util.h:74)
==1== by 0x140F97: manager_set_default_rlimits (manager.c:2929)
==1== by 0x1303DA: manager_set_defaults (main.c:737)
==1== by 0x133A02: main (main.c:1718)
==1==
==1== 272 bytes in 17 blocks are definitely lost in loss record 2 of 6
==1== at 0x4C2BBCF: malloc (vg_replace_malloc.c:299)
==1== by 0x1E350E: memdup (alloc-util.c:34)
==1== by 0x135AFB: memdup_multiply (alloc-util.h:74)
==1== by 0x140F97: manager_set_default_rlimits (manager.c:2929)
==1== by 0x1303DA: manager_set_defaults (main.c:737)
==1== by 0x13480D: main (main.c:1828)
==1==
==1== LEAK SUMMARY:
==1== definitely lost: 288 bytes in 18 blocks
==1== indirectly lost: 0 bytes in 0 blocks
==1== possibly lost: 0 bytes in 0 blocks
==1== still reachable: 61,440 bytes in 4 blocks
==1== suppressed: 0 bytes in 0 blocks
==1== Reachable blocks (those to which a pointer was found) are not
shown.
==1== To see them, rerun with: --leak-check=full --show-leak-kinds=all
When we are running in test mode, we don't expect any signals.
In fact ^C should end the program. This also avoids permission
issues when running systemd-analyze verify.
When the daemon reloads, it doesn not actually give up its DBus connection,
as wrongly stated in an earlier commit. However, even though the bus
connection stays open, the daemon flushes out all its internal state.
Hence, if there is a NameOwnerChanged signal after the flush and before the
deserialization, it cannot be matched against any pending unit.
To fix this, rename bus_list_names() to manager_sync_bus_names() and call
it explicitly at the end of the daemon reload operation.
Output the same message when a request to change the log level is
received over dbus and through a signal. From the user point of view
those two operations are very similar and it's easy to think that the
dbus operation didn't work when the expected message is not emitted.
Also "downgrade" the message level to info, since this is a normal
user initiated action.
This way we can only print the debug message when the status actually
changes. We also means we don't print anything when running in --user
mode, where status output is always disabled.
GLIB has recently started to officially support the gcc cleanup
attribute in its public API, hence let's do the same for our APIs.
With this patch we'll define an xyz_unrefp() call for each public
xyz_unref() call, to make it easy to use inside a
__attribute__((cleanup())) expression. Then, all code is ported over to
make use of this.
The new calls are also documented in the man pages, with examples how to
use them (well, I only added docs where the _unref() call itself already
had docs, and the examples, only cover sd_bus_unrefp() and
sd_event_unrefp()).
This also renames sd_lldp_free() to sd_lldp_unref(), since that's how we
tend to call our destructors these days.
Note that this defines no public macro that wraps gcc's attribute and
makes it easier to use. While I think it's our duty in the library to
make our stuff easy to use, I figure it's not our duty to make gcc's own
features easy to use on its own. Most likely, client code which wants to
make use of this should define its own:
#define _cleanup_(function) __attribute__((cleanup(function)))
Or similar, to make the gcc feature easier to use.
Making this logic public has the benefit that we can remove three header
files whose only purpose was to define these functions internally.
See #2008.
Now that we don't have RequiresOverridable= and RequisiteOverridable=
dependencies anymore, we can get rid of tracking the "override" boolean
for jobs in the job engine, as it serves no purpose anymore.
While we are at it, fix some error messages we print when invoking
functions that take the override parameter.
We can't handle errors of thisc all sanely anyway, and we never actually
return any errors from the unit type that implements the call. Hence,
let's make this void, in order to simplify things.
We cannot handle enumeration failures in a sensible way, hence let's try
hard to continue without making such failures fatal, and log about it
with precise error messages.
Don't simply continue after sleeping, it potentially puts us in a state
of spinning doing nothing slowly, if the ratelimit_test() keeps
detecting the need for limiting.
Observed in vms after the host had been suspended for a while, on resume
systemd entered a loop of making zero progress spamming the console
with:
[431942.850090] systemd[1]: Looping too fast. Throttling execution a
little.
I see no reason to have a continue here, the intention should be to
throttle execution, not circumvent it altogether.
Let's make sure we don't start blocking on sd_notify() earlier than
necessary, let's bump the socket buffer sizes to 8M.
We already do something similar for our logging socket buffers, hence
apply a similar bump here.
Currently, we dispatch NOTIFY messages in a tight loop. Regardless how
much data is incoming, we always dispatch everything that is queued.
This, however, completely breaks priority event-handling of sd-event.
When dispatching one NOTIFY event, another completely different event
might fire, or might be queued by the NOTIFY handling. However, this
event will not get dispatched until all other further NOTIFY messages are
handled. Those might even arrive _after_ the other event fired, and as
such completely break priority ordering of sd-event (which several code
paths rely on).
Break this by never dispatching multiple messages. Just return after each
message that was read and let sd-event handle everything else.
(The patch looks scarier that it is. It basically just drops the for(;;)
loop and re-indents the loop-content.)
There are more than enough calls doing string manipulations to deserve
its own files, hence do something about it.
This patch also sorts the #include blocks of all files that needed to be
updated, according to the sorting suggestions from CODING_STYLE. Since
pretty much every file needs our string manipulation functions this
effectively means that most files have sorted #include blocks now.
Also touches a few unrelated include files.
This adds support for naming file descriptors passed using socket
activation. The names are passed in a new $LISTEN_FDNAMES= environment
variable, that matches the existign $LISTEN_FDS= one and contains a
colon-separated list of names.
This also adds support for naming fds submitted to the per-service fd
store using FDNAME= in the sd_notify() message.
This also adds a new FileDescriptorName= setting for socket unit files
to set the name for fds created by socket units.
This also adds a new call sd_listen_fds_with_names(), that is similar to
sd_listen_fds(), but also returns the names of the fds.
systemd-activate gained the new --fdname= switch to specify a name for
testing socket activation.
This is based on #1247 by Maciej Wereski.
Fixes#1247.
Let's underline the header line of the table shown by cgtop, how it is
customary for tables. In order to do this, let's introduce new ANSI
underline macros, and clean up the existing ones as side effect.
Add a new config directive called NetClass= to CGroup enabled units.
Allowed values are positive numbers for fix assignments and "auto" for
picking a free value automatically, for which we need to keep track of
dynamically assigned net class IDs of units. Introduce a hash table for
this, and also record the last ID that was given out, so the allocator
can start its search for the next 'hole' from there. This could
eventually be optimized with something like an irb.
The class IDs up to 65536 are considered reserved and won't be
assigned automatically by systemd. This barrier can be made a config
directive in the future.
Values set in unit files are stored in the CGroupContext of the
unit and considered read-only. The actually assigned number (which
may have been chosen dynamically) is stored in the unit itself and
is guaranteed to remain stable as long as the unit is active.
In the CGroup controller, set the configured CGroup net class to
net_cls.classid. Multiple unit may share the same net class ID,
and those which do are linked together.
The current implementation directly monitor /proc/self/mountinfo and
/run/mount/utab files. It's really not optimal because utab file is
private libmount stuff without any official guaranteed semantic.
The libmount since v2.26 provides API to monitor mount kernel &
userspace changes and since v2.27 the monitor is usable for
non-root users too.
This patch replaces the current implementation with libmount based
solution.
Signed-off-by: Karel Zak <kzak@redhat.com>
Let's make sure that we follow the same codepaths when adjusting a
cgroup property via the dbus SetProperty() call, and when we execute the
StartupCPUShares= effect.
Introduce a proper enum, and don't pass around string ids anymore. This
simplifies things quite a bit, and makes virtualization detection more
similar to architecture detection.
Let's move the actual cgroup part of it into a new separate function
manager_get_unit_by_pid_cgroup(), and then make
manager_get_unit_by_pid() just a wrapper that also checks the two pid
hashmaps.
Then, let's make sure the various calls that want to deliver events to
the owners of a PID check both hashmaps and the cgroup and deliver the
event to *each* of them. OTOH make sure bus calls like GetUnitByPID()
continue to check the PID hashmaps first and the cgroup only as
fallback.
This adds a new PID_TO_PTR() macro, plus PTR_TO_PID() and makes use of
it wherever we maintain processes in a hash table. Previously we
sometimes used LONG_TO_PTR() and other times ULONG_TO_PTR() for that,
hence let's make this more explicit and clean up things.
This patch set adds full support the new unified cgroup hierarchy logic
of modern kernels.
A new kernel command line option "systemd.unified_cgroup_hierarchy=1" is
added. If specified the unified hierarchy is mounted to /sys/fs/cgroup
instead of a tmpfs. No further hierarchies are mounted. The kernel
command line option defaults to off. We can turn it on by default as
soon as the kernel's APIs regarding this are stabilized (but even then
downstream distros might want to turn this off, as this will break any
tools that access cgroupfs directly).
It is possibly to choose for each boot individually whether the unified
or the legacy hierarchy is used. nspawn will by default provide the
legacy hierarchy to containers if the host is using it, and the unified
otherwise. However it is possible to run containers with the unified
hierarchy on a legacy host and vice versa, by setting the
$UNIFIED_CGROUP_HIERARCHY environment variable for nspawn to 1 or 0,
respectively.
The unified hierarchy provides reliable cgroup empty notifications for
the first time, via inotify. To make use of this we maintain one
manager-wide inotify fd, and each cgroup to it.
This patch also removes cg_delete() which is unused now.
On kernel 4.2 only the "memory" controller is compatible with the
unified hierarchy, hence that's the only controller systemd exposes when
booted in unified heirarchy mode.
This introduces a new enum for enumerating supported controllers, plus a
related enum for the mask bits mapping to it. The core is changed to
make use of this everywhere.
This moves PID 1 into a new "init.scope" implicit scope unit in the root
slice. This is necessary since on the unified hierarchy cgroups may
either contain subgroups or processes but not both. PID 1 hence has to
move out of the root cgroup (strictly speaking the root cgroup is the
only one where processes and subgroups are still allowed, but in order
to support containers nicey, we move PID 1 into the new scope in all
cases.) This new unit is also used on legacy hierarchy setups. It's
actually pretty useful on all systems, as it can then be used to filter
journal messages coming from PID 1, and so on.
The root slice ("-.slice") is now implicitly created and started (and
does not require a unit file on disk anymore), since
that's where "init.scope" is located and the slice needs to be started
before the scope can.
To check whether we are in unified or legacy hierarchy mode we use
statfs() on /sys/fs/cgroup. If the .f_type field reports tmpfs we are in
legacy mode, if it reports cgroupfs we are in unified mode.
This patch set carefuly makes sure that cgls and cgtop continue to work
as desired.
When invoking nspawn as a service it will implicitly create two
subcgroups in the cgroup it is using, one to move the nspawn process
into, the other to move the actual container processes into. This is
done because of the requirement that cgroups may either contain
processes or other subgroups.
Zbigniew Jędrzejewski-Szmek