This compares two PidRef structures via the pid_t field. Ideally we'd do
a stricter comparison here, that is safe towards PID reuse, but so far
the pidfd API lacks suitable mechanisms for that, hence do the best we
can do.
The kernel loads the DTB from EFI before ExitBootServices():
https://github.com/torvalds/linux/blob/v6.5/drivers/firmware/efi/libstub/fdt.c#L245
DTBs can map and assign arbitrary memory ranges. The kernel refuses
to load one from the dtb= kernel command line parameter when secure
boot is enabled, as it's not safe. Let's do the same for type 1
entries, as they are unverified.
This only affects arm64 and riscv64, firmwares do not support DTB
on x86.
Let's not stop watching processes on every scope state change. This
corrects behaviour when a scope is being enqueued and hasn't started
yet, but has processes associated to it already. previously, if we'd doa
full PID 1 reload we'd stop watching those pids. With this change we'll
continue watching them in that case, and only stop watching them when
the scope unit really shuts down after first being up.
We only need a separate mount namespace if we're operating on a
btrfs block device so let's make sure we only unshare the mount
namespace if that's the case.
Replaces #29214
The implementations are not 100% overlapping, so use different identifiers, so
that revocations can be done independently. e.g.: a bug that affects only
sd-boot won't necessarily cause old UKIs to be revoked.
The lines in kbd-model-map date back to ye olde times (RH's old
system-config-keyboard), and I think predate this bug:
https://bugzilla.redhat.com/show_bug.cgi?id=1039185
where we got strong feedback that, for 'switched' layout setups
like Russian, US English should be the *first* layout and the
native layout the *second* one. This is how anaconda and, as of
recently, gnome-initial-setup configure such cases - but that
means, if we try to use localed to convert these configurations
using kbd-model-map, we get the wrong result (we get "us" as the
console layout). See also:
https://bugzilla.redhat.com/show_bug.cgi?id=1912609
where we first noticed this wasn't working right, but sadly, we
'fixed' it with a not-really-correct bodge in anaconda instead
of doing it properly.
Signed-off-by: Adam Williamson <awilliam@redhat.com>
We should give a match bonus if the X context variant is empty
and the xvariant column in kbd-model-map is "-" (which means
none). Currently, we don't, which means that if you call this
on a context with layouts bg,us and no variant, you get the
console layout bg_pho-utf8 instead of bg_bds-utf8 (because both
score the same, and the bg_pho-utf8 row comes first). You should
get bg_bds-utf8 in this case.
Signed-off-by: Adam Williamson <awilliam@redhat.com>
This module is only available on PPC hw, so avoid trying to load it elsewhere, as it generates a misleading error message in the logs:
modprobe: FATAL: Module tpm_ibmvtpm not found in directory /lib/modules/5.15.0-83-generic
As I noticed a lot of missing information when trying to implement checking
for missing info. I reimplemented the version information script to be more
robust, and here is the result.
Follow up to ec07c3c80b
This ports over unit_main_pid() + unit_control_pid() to return PidRef*
pointers (which also means the underlying UnitVTable function pointers
are changed accordingly).
This then uses te functions to simplify the unit_kill() call, by
avoiding the kill() vtable indirection and instead just suing
unit_main_pid() and unit_control_pid() directly.
This adds a new "PollLimit" pair of settings to .socket units, very
similar to existing "TriggerLimit" logic. The differences are:
* PollLimit focusses on the polling on the sockets, and pauses that
temporarily if a ratelimit on that is reached. TriggerLimit otoh
focusses on the triggering effect of socket units, and stops
triggering once the ratelimit is hit.
* While the trigger limit being hit is an action that causes the socket
unit to fail the polling limit being reached will just temporarily
disable polling on the socket fd, and it is resumed once the ratelimit
interval is over.
* When a socket unit operates on multiple socket fds (e,g, ListenStream=
on both some ipv6 and an ipv4 address or so). Then the PollLimit will
be specific to each fd, while the trigger limit is specific to the
whole unit.
Implementation-wise this is mostly a wrapper around sd-event's
sd_event_source_set_ratelimit(), which exposes the desired behaviour
directly.
Usecase for all of this: socket services which when overloaded with
connections should just slow down reception of it, but not fail
persistently.
Let's tweak the message if not enough swap is around slightly: systems
might have plenty swap backed by incompatible storage (specifically:
swap files on btrfs), but we (currently) do not support hibernating to
that.
Hence let's say *suitable* swap space and talk about *compatibility* of
backing storage.
Hopefully this will make things a bit clearer to users.
Prompted by: #29189
As pointed out in the review, all this applies to the user services too, so are
not managed by the "init system", but by the more generic "service manager".
Also:
- use oxford comma
- change "employ" to "use" in various places
- change "the init system forwards messages to syslog" to "are forwarded to
syslog". This is done by systemd-journald, so really there is no forwarding,
because systemd-journald just writes them to a file in the common setup,
so let's use the passive form to avoid specifying who does this.
This conceptually reverts e95acdfe1d,
but the actual contents of the script are taken from the command invocation
in meson with all the updates that happened in the meantime.
One small change is that I replaced () by {}: this avoids one subprocess spawn.
People were worried about the cost of vcs_tag(), and this microoptimization may
help a bit. I measured the speed on machine, and noop rebuilds are still about
100–120 ms.
The logic is entirely moved to the script. This makes the meson config simpler
and also makes it easier to use it externally.
The script is needed for in-place rpm builds, see README.build-in-place.md [1],
where it is invoked from the spec file to determine the project version.
[1] https://src.fedoraproject.org/rpms/systemd/blob/rawhide/f/README.build-in-place.md
Lennart Poettering
Mike Yuan
Luca Boccassi
Daan De Meyer
IllusionMan1212
Daniel Rusek
Zbigniew Jędrzejewski-Szmek
Abderrahim Kitouni
Joerg Behrmann
Adam Williamson
Valentin David
наб
Dan Streetman