Without this change, there are no ordering between udevd and tmpfiles,
and if tmpfiles is invoked later it may discard the permission set by
udevd.
Fixes an issue introduced by b42482af90.
Fixes#28588 and #28653.
We are basically already there, just need to add MONOTONIC_USEC= to the
RELOADING=1 message, and make sure the message is generated in really
all cases.
This partially reverts cabc1c6d7a.
The setting ProtectClock= implies DeviceAllow=, which is not suitable
for udevd. Although we are slowly removing cgropsv1 support, but
DeviceAllow= with cgroupsv1 is necessarily racy, and reloading PID1
during the early boot process may cause issues like #24668.
Let's disable ProtectClock= for udevd. And, if necessary, let's
explicitly drop CAP_SYS_TIME and CAP_WAKE_ALARM (and possibly others)
by using CapabilityBoundingSet= later.
Fixes#24668.
And enable cgroup delegation for udevd.
Then, processes invoked through ExecReload= are assigned .control
subcgroup, and they are not killed by cg_kill().
Fixes#16867 and #22686.
Programs run by udev triggers may need to execute the bpf() syscall. Even more
so, since on a cgroup v2 system, the only way to set up device access filtering
is to install a BPF program on the cgroup in question and one way of passing
data to such program is through BPF maps, which can only be access using the
bpf() syscall. One such use case was identified in RHBZ#2025264 related to
snap-device-helper, and led to RHBZ#2027627 being filed.
Unfortunately there is no finer grained control over what gets passed in the
syscall, so just enable bpf() and leave fine grained mediation to other
security layers (eg. SELinux).
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=2027627
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
We don't need two (and half) templating systems anymore, yay!
I'm keeping the changes minimal, to make the diff manageable. Some enhancements
due to a better templating system might be possible in the future.
For handling of '## ' — see the next commit.
Add `ProtectClock=yes` to systemd units. Since it implies certain
`DeviceAllow=` rules, make sure that the units have `DeviceAllow=` rules so
they are still able to access other devices. Exclude timesyncd and timedated.
This makes things a bit simpler and the build a bit faster, because we don't
have to rewrite files to do the trivial substitution. @rootbindir@ is always in
our internal $PATH that we use for non-absolute paths, so there should be no
functional change.
As discussed on systemd-devel [1], in Fedora we get lots of abrt reports
about the watchdog firing [2], but 100% of them seem to be caused by resource
starvation in the machine, and never actual deadlocks in the services being
monitored. Killing the services not only does not improve anything, but it
makes the resource starvation worse, because the service needs cycles to restart,
and coredump processing is also fairly expensive. This adds a configuration option
to allow the value to be changed. If the setting is not set, there is no change.
My plan is to set it to some ridiculusly high value, maybe 1h, to catch cases
where a service is actually hanging.
[1] https://lists.freedesktop.org/archives/systemd-devel/2019-October/043618.html
[2] https://bugzilla.redhat.com/show_bug.cgi?id=1300212
If logging disappears issues are hard to debug, hence let's give
journald a slight edge over other services when the OOM killer hits.
Here are the special adjustments we now make:
systemd-coredump@.service.in OOMScoreAdjust=500
systemd-journald.service.in OOMScoreAdjust=-250
systemd-udevd.service.in OOMScoreAdjust=-1000
(i.e. the coredump processing is made more likely to be killed on OOM,
and udevd and journald are less likely to be killed)
This is generally the safer approach, and is what container managers
(including nspawn) do, hence let's move to this too for our own
services. This is particularly useful as this this means the new
@system-service system call filter group will get serious real-life
testing quickly.
This also switches from firing SIGSYS on unexpected syscalls to
returning EPERM. This would have probably been a better default anyway,
but it's hard to change that these days. When whitelisting system calls
SIGSYS is highly problematic as system calls that are newly introduced
to Linux become minefields for services otherwise.
Note that this enables a system call filter for udev for the first time,
and will block @clock, @mount and @swap from it. Some downstream
distributions might want to revert this locally if they want to permit
unsafe operations on udev rules, but in general this shiuld be mostly
safe, as we already set MountFlags=shared for udevd, hence at least
@mount won't change anything.
Since hotplugs happen as soon as udevd is started, there is not much sense
in giving udev-trigger an After= dependency on any service. The device
could be hotplugged before coldplug starts.
This is intended to avoid the race window where we create the hwdb with
the wrong selinux context (then fix it up afterwards).
https://github.com/systemd/systemd/issues/3458#issuecomment-322444107
Let's make this an excercise in dogfooding: let's turn on more security
features for all our long-running services.
Specifically:
- Turn on RestrictRealtime=yes for all of them
- Turn on ProtectKernelTunables=yes and ProtectControlGroups=yes for most of
them
- Turn on RestrictAddressFamilies= for all of them, but different sets of
address families for each
Also, always order settings in the unit files, that the various sandboxing
features are close together.
Add a couple of missing, older settings for a numbre of unit files.
Note that this change turns off AF_INET/AF_INET6 from udevd, thus effectively
turning of networking from udev rule commands. Since this might break stuff
(that is already broken I'd argue) this is documented in NEWS.
udevd already limits its number of workers/children: the max number is actually
twice the number of CPUs the system is using.
(The limit can also be raised with udev.children-max= kernel command line
option BTW).
On some servers, this limit can easily exceed the maximum number of tasks that
systemd put on all services, which is 512 by default.
Since udevd has already its limitation logic, simply disable the static
limitation done by TasksMax.
Apparently, disk IO issues are more frequent than we hope, and 1min
waiting for disk IO happens, so let's increase the watchdog timeout a
bit, for all our services.
See #1353 for an example where this triggers.
ReadOnlySystem= uses fs namespaces to mount /usr and /boot read-only for
a service.
ProtectedHome= uses fs namespaces to mount /home and /run/user
inaccessible or read-only for a service.
This patch also enables these settings for all our long-running services.
Together they should be good building block for a minimal service
sandbox, removing the ability for services to modify the operating
system or access the user's private data.
As of kmod v14, it is possible to export the static node information from
/lib/modules/`uname -r`/modules.devname in tmpfiles.d(5) format.
Use this functionality to let systemd-tmpfilesd create the static device nodes
at boot, and drop the functionality from systemd-udevd.
As an effect of this we can move from systemd-udevd to systemd-tmpfiles-setup-dev:
* the conditional CAP_MKNOD (replaced by checking if /sys is mounted rw)
* ordering before local-fs-pre.target (see 89d09e1b5c)
Not that it would matter much, but let's make things a bit more
systematic: early boot services shall order themselves before
sysinit.target, and nothing else.
static nodes (like /dev/loop-control) are created when systemd-udevd
is started and needed to mount loopback devices. Therefore,
local-fs-pre.target should be only started after systemd-udevd is
started.
This naming convention is more inline with other systemd daemon
unit names (systemd-logind.service, systemd-localed.service etc)
The companion .socket units have also been renamed, however the
-trigger and -settle units keep their current name as these are
not directly related to daemon process itself.
