When restoring the COW flag for journals on BTRFS, the full journal contents
are copied into new files. But during these operations, the acls of the
previous files were lost and users were not able to access to their old
journal contents anymore.
It's not clear why we rescheduled a service auto restart while a stop job for
the unit was pending. The comment claims that the unit shouldn't be restarted
but the code did reschedule an auto restart meanwhile.
In practice that was rarely an issue because the service waited for the next
auto restart to be rescheduled, letting the queued stop job to be proceed and
service_stop() to be called preventing the next restart to complete.
However when RestartSec=0, the timer expired right away making PID1 to
reschedule the unit again, making the timer expired right away... and so
on. This busy loop prevented PID1 to handle any queued jobs (and hence giving
no chance to the start rate limiting to trigger), which made the busy loop last
forever.
This patch breaks this loop by skipping the reschedule of the unit auto restart
and hence not depending on the value of u->restart_usec anymore.
Fixes: #13667
Writing a byte to test10.socket is actually the root cause of issue #19154:
depending on the timing, it's possible that PID1 closes the socket before socat
(or nc, it doesn't matter which tool is actually used) tries to write that one
byte to the socket. In this case writing to the socket returns EPIPE, which
causes socat to exit(1) and subsequently make the test fail.
Since we're only interested in connecting to the socket and triggering the rate
limit of the socket, this patch removes the parts that write the single byte to
the socket, which should remove the race for good.
Since it shouldn't matter whether the test uses socat or nc, let's switch back
to nc and hence remove the sole user of socat. The exit status of nc is however
ignored because some versions might choke when the socket is closed
unexpectedly.
Due to the fact that systemd-journal-flush.service has
"Requires=systemd-journald.service", this service is stopped too when journald
is requested to do so.
However stopping systemd-journal-flush.service implies that journald
relinquishes /var hence implicitly switching back to the volatile storage
mode and removing /run/systemd/journal/flushed.
If journald is started afterwards, it will run in volatile storage mode
regardless of the value of 'Storage=' as it believes now that /var is not yet
ready (because the flushed flag is missing).
Because this flag is mainly an indication for journald that the initialization
of /var/log/journal (during the boot process) has been done,
systemd-journal-flush.service shouldn't be tied to the state of journald itself
but to the state of /var/log/journal, hence to the state of the system.
Follow-up for 7d85383edb.
Apparently the previous limit set on the max number of inodes for /dev was too
small as a system with 4096 LUNs attached can consume up to 95k inodes for
symlinks:
# /bin/df -i
Filesystem Inodes IUsed IFree IUse% Mounted on
devtmpfs 49274377 95075 49179302 1% /dev
Hence this patch bumps the limit from 64k to 128k although the new limit is
still pretty arbitrary (that said, not sure if it really makes sense to put
such absolute limit number).
Forcing a specific fs for the image is usually a bad idea because the initrd
(borrowed from the host) is likely to include only support for the filesystem
used by the host's rootfs.
Since the point of this test is to check aliases on mount units, there's no
specific need for ext4, hence drop any parts that request or rely on ext4.
This test makes assumptions on the availability of some mappings contained in
kbd-model-map and therefore strongly relies on the version shipped by
upstream. IOW the test is likely to fail if it's installed on a system with a
more comprehensive kbd-model-map.
This patch makes the upstream kbd-model-map file available via a symlink in
test/testdata/test-keymap-util dir and makes sure that this specific version is
always used by test-keymap-util regardless of whether the test is installed and
run on a different system or directly run (optionally via meson) from the
project working dir.
If an error happened while the timeout value was being programmed, the error
was ignored and the watchdog module used the new timeout value whereas the
watchdog device was left with the previous one.
Now in cases of error, the device is now disabled and closed if it wasn't
opened already otherwise the previous timeout value is kept so the device is
still pinged at correct intervals.
Passing 0 meant "disable the watchdog although still kept it opened". However
this case didn't seem to be useful especially since PID1 closes the device if
it is passed the nul timeout.
Hence let's change the meaning of watchdog_setup(0) to match PID1's behavior
which allows to simplify the code a bit.
Hence this patch also drops enable_watchdog().
This patch changes the meaning of USEC_INFINITY value for the watchdog
module.
Previously passing this value was a NOP. It now has a special meaning: it
requests the watchdog module to read the programmed timeout value and reuse it
for pinging the device.
This is mostly useful when the watchdog is started by the firmware and there's
no way to reconfigure the timeout with a different value
afterwards. "RuntimeWatchdogSec=infinity" in system.conf can now be used rather
than putting an arbitrary value that PID1 will fail to set (even if it still
felt back to the programmed timeout).
Please note that "infinity" is not supposed to restore the default value of the
firmware. If the value is changed after booting then "infinity" would simply
reuse the current programmed value. IOW it's a NOP unless the watchdog was
previously closed and in that case it will be reopened and the last programmed
value reused.
The second call to name_to_handle_at_loop() didn't check for the specific
errors that can happen when the parent dir is mounted by nfs and instead of
falling back like it's done for the child dir, fd_is_mount_point() failed in
this case.
Even if journald was not running, flushing the volatile journal used to work if
the journal was already flushed (ie the flushed flag
/run/systemd/journald/flushed was created).
However since commit 4f413af2a0, this behavior changed and now '--flush' fails
because it tries to contact journald without checking the presence of the
flushed flag anymore.
This patch restores the previous behavior since there's no reason to fail when
journalctl can figure out that the flush is not necessary.
Store the actual timeout value found by the driver in watchdog_timeout since
this value is more accurate for calculating the next time for pinging the
device.
This basically reverts commit 61927b9f11 and
relies on the fact that watchdog_ping() will open and setup the watchdog for us
in case the device appears later on.
Also unlike what is said in comment
https://github.com/systemd/systemd/pull/17460#pullrequestreview-517434377, both
m->watchdog[] and m->overriden_watchdog[] are not supposed to store the actual
timeout used by the watchdog device but stores the value defined by the user.
If the HW timeout value is really needed by the manager then it's probably
better to read it via an helper defined in watchdog.c instead. However the HW
timeout value is currently only needed by the watchdog code itself mainly when
it calculates the time for the next ping.
This patch allows watchdog_ping() to be used unconditionally regardless of
whether watchdog_set_timeout() or watchdog_close() has been previously called
or not and in both cases watchdog_ping() does nothing.
shutdown.c has been updated to cope with this change.
This reverts commit 491b736a49.
If the _static_ linked version of busybox is installed, openSUSE doesn't need
any specific code.
A following commit will make sure that the static linked version of busybox is
installed in the busybox container.
NO_BUILD=1 indicates that we want to test systemd from the local system and not
the one from the local build. Hence there should be no need to call
find-build-dir.sh when NO_BUID=1 especially since it's likely that the script
will fail to find a local build in this case.
This avoids find-build-dir.sh to emit 'Specify build directory with $BUILD_DIR'
message when NO_BUILD=1 and no local build can be found.
This introduces a behavior change though: systemd from the local system will
always be preferred when NO_BUILD=1 even if a local build can be found.
Before this patch, there was no way to request all running user instances for
reexecuting. However this can be useful especially during package updates
otherwise user instances are never updated and keep running a potentially very
old version of the binaries.
Now assuming that we have enough priviledge, it's possible to request
reexecution of all user instances:
systemctl kill --signal=SIGRTMIN+25 "user@*.service"
Note that this request is obviously asynchronous as it relies on a
signal. Keeping "systemctl kill" as the only interface should be good enough to
make this obvious and that's the reason why another interface, such as
"systemctl --global daemon-reexec" has not been considered.
PID1 already uses SIGTERM for reexecuting hence sending it SIGRTMIN+25 is a
nop.
This graphic chip doesn't have a DRM driver and fall back to vesa-framebuffer
driver.
Without this patch, users of such chip suddenly see their GUI broken without
any indication or reason of what happened (no error message). Hence this
regression is near to impossible to troubleshoot for end users. Such case was
reported https://bugzilla.opensuse.org/show_bug.cgi?id=1187154.
Rather than adding another exception in the udev rules to deal with such
HWs, they instead get their own hwdb file '60-seat.hwdb'.
This graphic chip doesn't have a DRM driver and fallback to vesa-framebuffer
driver.
Without this patch, users of such chip suddenly see their GUI broken without
any indication or reason of what happened (no error message). Hence this
regression is near to impossible to troubleshoot for end users.
Commit 428a9f6f1d freed u->pids which is
problematic since the references to this unit in m->watch_pids were no more
removed when the unit was freed.
This patch makes sure to clean all this refs up before freeing u->pids by
calling unit_unwatch_all_pids().
Commit 42cc2855ba incorrectly removed the condition on sysfs in both
sys-fs-fuse-connections.mount and sys-kernel-config.mount. However there are
still needed in case modprobe of one of these modules is intentionally skipped
(due to lack of privs for example).
This patch restores the 2 conditions which should be safe for the common case,
since all conditions are only checked after all deps ordered before are
complete.
Follow-up for 42cc2855ba.
Otherwise if a daemon-reload happens somewhere between the enqueue of the job
start for the scope unit and scope_start() then u->pids might be lost and none
of the processes specified by "PIDs=" will be moved into the scope cgroup.
udev requests to start the fs mount units when their respective module is
loaded. For that it monitors uevents of type "ADD" for the relevant fs modules.
However the uevent is sent by the kernel too early, ie before the init() of the
module is called hence before directories in /sys/fs/ are created.
This patch workarounds adds "Requires/After=modprobe@<fs-module>.service" to
the mount unit, which means that modprobe(8) will be called once the fs module
is announced to be loaded. This sounds pointless, but given that modprobe only
returns after the initialization of the module is complete, it should
workaround the issue.
As a side effect, the module will be automatically loaded if the mount unit is
started manually.
Fixes#17586.
This reverts commit 9cbf1e58f9.
The presence of /sys/module/%I directory can't be used to assert that the load
of a given module is complete and therefore the call to modprobe(8) can be
skipped. Indeed this directory is created before the init() function of the
module is called.
Users of modprobe@.service needs to be sure that once this service returns the
module is fully operational.
This was needed before commit 16e4fd87c5 added a
mode that opens the log fds for every single log message. This mode is used in
execute.c since then making the explicit call to log_open unnecessary.
This basically reverts ea89a119cd.
Since commit 883eb9be98, vconsole-setup might be
called again to operate on dummy console where font operations are not
supported but where it's still important to have the correct keymap set [0][1].
vconsole-setup is mainly called by udev but can also be run via a dependency of
an early service. Both cases might end up calling vconsole-setup on the dummy
console.
The first case can happen during early boot even on systems that use (instead
of the dummy console) a "simple" video console driver supporting font
operations (such as vgacon) until a more specific driver (such as i915) takes
the console over. While this is happening vgacon is deactivated and temporarly
replaced by the dummy console [2].
There are also other cases where systemd-vconsole-setup might be called on
dummy console especially during (very) early boot. Indeed
systemd-vconsole-setup.service might be pulled in by early interactive services
such as 'dracut-cmdline-ask.service` which is run before udev.
If that happens on platforms with no grapical HWs (such as embedded ARM) or
with dummy console initially installed until a driver takes over (like Xen and
xen-fbfront) then setting font will fail.
Therefore this patch downgrades the log message emitted when setting font fails
to LOG_DEBUG and when font operations is not implemented like it's the case for
the dummy console.
Fixes: #16406.
[0] https://github.com/systemd/systemd/issues/10826
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1652473
[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/vga/vgaarb.c?h=v5.7#n204
If an entry in fstab uses "x-systemd.automount" option and also asks for
additionnal dependencies via x-systemd.requires or such, then the dependencies
were applied to the automount unit.
But this unlikely to do the right thing and is inconsistent with what's done
for network mounts.
Indeed when an fstab entries has "_netdev,x-systemd.automount" options, the
dependencies against the network requested by "_netdev" are (correctly) applied
to the mount unit only and the automount unit remains ordered against
local-fs.target.
The same logic should be followed when extra deps are specified via the mount
options as automount units should always be ordered against local-fs.target.
Note: in general explicit deps specified via mount options should be used with
care and should be used to specify dependencies on other mount units only as it
can easily create ordering cycles otherwise like it's been seen in
https://github.com/systemd/systemd-stable/issues/69. Mount units (as well as
automount ones) are ordered before local-fs.target by default which is a
low-level target that most other units depend on.
The only way to control "ShowStatus" property programmatically was to use the
signal API and wait until the property "ShowStatus" switched to the new value.
This interface is rather cumbersome to use and doesn't allow to temporarily
override the current setting and later restore the overridden value in
race-free manner.
The new method also accepts the empty string as argument which allows to
restore the initial value of ShowStatus, ie the value before it was overridden
by this method.
Fixes: #11447.
The fact that m->show_status was serialized/deserialized made impossible any
further customisation of this setting via system.conf. IOW the value was
basically always locked unless it was changed via signals.
This patch reworks the handling of m->show_status but also makes sure that if a
new value was changed via the signal API then this value is kept and preserved
accross PID1 reexecuting or reloading.
Note: this effectively means that once the value is set via the signal
interface, it can be changed again only through the signal API.
The name 'manager_get_show_status()' suggests that the function simply reads
the property 'show_status' of the manager and hence returns a 'StatusType'
value.
However it was doing more than that since it contained the logic (based on
'show_status' but also on the state of the manager) to figure out if status
message could be emitted to the console.
Hence this patch renames the function to 'manager_should_show_status()'. The
previous name will be reused in a later patch to effectively return the value
of 'show_status' property.
No functional change.
That's reduce the number of functions dealing with configuration
parsing/loading and should make the code simpler especially since this function
was used only once.
No functional change.
Most complexity of this patch is due to the fact that some manager settings
(basically the watchdog properties) can be set at runtime and in this case the
runtime values must be retained over daemon-reload or daemon-reexec.
For consistency sake, all watchdog properties behaves now the same way, that
is:
- Values defined by config files can be overridden by writing the new value
through their respective D-BUS properties. In this case, these values are
preserved over reload/reexec until the special value '0' or USEC_INFINITY
is written, which will then restore the last values loaded from the config
files. If the restored value is '0' or 'USEC_INFINITY', the watchdogs will
be disabled and the corresponding device will be closed.
- Reading the properties from a user instance will return the USEC_INFINITY
value as these properties are only meaningful for PID1.
- Writing to one of the watchdog properties of a user instance's will be a
NOP.
Fixes: #15453
Regardless of whether a mount is setup in initrd or int the main system,
the network default dependencies _netdev should still be honored.
IOW if a mount unit use the following options "x-initrd.mount,_netdev", it
should be ordered against initrd-fs.target, network.target,
network-online.target.
/dev/vdb1 /mnt ext4 x-initrd.mount,_netdev defaults 0 0
Before this patch:
Before=umount.target initrd-fs.target
After=system.slice sysroot.mount dev-vdb1.device -.mount systemd-journald.socket blockdev@dev-vdb1.target
After this patch:
Before=initrd-fs.target umount.target
After=network-online.target -.mount blockdev@dev-vdb1.target dev-vdb1.device sysroot.mount system.slice network.target systemd-journald.socket
First After=local-fs-pre.target wasn't described in the man page although it's
part of the default dependencies automatically set by pid1.
Secondly, Before=local-fs.target was only set if the automount unit was
generated from the fstab-generator because the dep was explicitly
generated. It was also not documented as a default dependency.
Fix it by managing the dep from pid1 instead.
fstab-generator was also handling the default ordering dependencies for mount
units setup in initrd. To do that it was turning the defaults dependencies off
completely and ordered the mount unit against either local-fs.target or
initrd-fs.target or initrd-root-fs.target itself.
But it had the bad side effect to also remove all other default dependencies as
well. Thus if an initrd mount was using _netdev, the network dependencies were
missing.
In general fstab-generator shouldn't use DefaultDependecies=no because it can
handle only a small set of the default dependencies the rest are dealt by pid1.
So this patch makes pid1 handle all default dependencies.
This patch changes the way user managers set the default umask for the units it
manages.
Indeed one can expect that if user manager's umask is redefined through PAM
(via /etc/login.defs or pam_umask), all its children including the units it
spawns have their umask set to the new value.
Hence make user units inherit their umask value from their parent instead of
the hard coded value 0022 but allow them to override this value via their unit
file.
Note that reexecuting managers with 'systemctl daemon-reexec' after changing
UMask= has no effect. To take effect managers need to be restarted with
'systemct restart' instead. This behavior was already present before this
patch.
Fixes#6077.
Both fstab-generator and pid1 are duplicating the handling of
"Before=local-fs.target" dependency for mount units.
fstab-generator is correctly skipping this dep if the mount unit is activated
by an automount unit.
However the condition used by pid1 was incorrect and missed the case when a mount
unit uses "x-systemd.automount" since in this case the mount unit should be
(only) ordered against its automount unit counterpart instead.
Stop generating device dependencies for extrinsic mounts: we already exclude
extrinsic mounts from the usual start-up and shutdown dependencies but some
extra deps added by generator_write_device_deps() were remaining.
Scheduling devices after a given unit can be useful to start device *jobs* at a
specific time in the transaction, see commit 4195077ab4.
This (hidden) change was introduced by commit eef85c4a3f.
Franck Bui