docs: invite people to define GPT partition types for all archs now
The discovery partitions spec so far suggested we should define
arch-specific partition type uuids only for archs that have EFI. Let's
change that and invite people to define them for any arch. Why? Even if
GPT is defined as part of the UEFI spec it's quite useful independently
of it, too. Specifically, our image dissection logic makes use of it,
i.e. systemd-nspawn, systemd-gpt-auto-generator, systemd-repart,
RootImage=, portable services, and so on. None of these tools are
related to UEFI in any way.
Anita Zhang [Tue, 9 Nov 2021 23:26:28 +0000 (15:26 -0800)]
core: replace slice dependencies as they get added
Defines a "UNIT_DEPENDENCY_SLICE_PROPERTY" UnitDependencyMask type that
is used when adding slices to the dependencies hashmap. This type is
used to remove slice dependencies when they get overridden by new ones.
nspawn: don't muck with caps if no network setting is used in settings file
Our goal here (as in the previous commits) is to ensure that a settings
file loaded in --settings=override mode is truly a NOP. Previously this
was not the case as we'd drop CAP_NET_ADMIN from the caps if the
settings file didn't enable networking.
With this change we'll drop it only if explicitly turned off in the
settings file, and otherwise let the built-in defaults and cmdline
params reign supreme as documented.
nspawn: use three boolean fields from settings file when actually set
Let's turn these three fields into tristates, so that we can distinguish
whether they are not configured at all from explicitly turned off.
Let#s then use this to ensure that we only copy the settings fields into
our execution environment if they are actually configured.
We already do this for some of the boolean settings, this adds it for
the missing ones.
The goal here is to ensure that an empty settings file used in
--settings=override mode (i.e. the default mode used in the
systemd-nspawn@.service unit) is truly a NOP.
The new helper returns whether the settings file had *any* networking
setting configured at all. We already have a similar helper
settings_private_network() which returns a similar result. The
difference is that the new helper will return true when the private
network was explicitly turned off, while the old one will only return
true if configured and enabled.
We'll reuse the helper a 2nd time later on, but even without it it makes
things a bit more readable.
repart: fix free area calculations for unaligned partitions
To properly detect how much space we have to distribute we need to take
into account that both the partition offset and the partition size
aren't aligned.
repart: don't distribute space after unaligned partitions
If we operate on a disk that has a pre-existing unaligned partition
(i.e. one that doesn't start on multiple of 4K, or doesn't have a size
of multiple 4K), then the amount of space after it to distribute among
partitions isn't a multiple of 4K either. So far we might end up
passing the remaining fraction to any partition that wanted it, which
was usually the first one after it that is newly defined. This then
confused the later placement algorithm, since it assumed all partitions
we newly allocate were properly aligned but by being extended by the
fractional space they wouldn't be anymore.
Let's hence fix that by ensuring we never pass space to later partitions
so that things wouldn't be aligned anymore.
Anything that is left-over then at the very end (i.e. typically exactly
the remaining fraction) is added as padding to the existing, unaligned
partition, so that it can't confuse anyone.
sd-event: don't destroy inotify data structures from inotify event handler
This fixes a bad memory access when we destroy an inotify source handler
from the handler itself, and thus destroy the associated inotify_data
structures.
The image build failed during kernel RPM installation (bug in %post
scriptlet). This has been fixed in the package suse-module-tools 16.0.13
[1]. The fix is in openSUSE Tumbleweed repos so the tests can be enabled
again.
Yu Watanabe [Sat, 6 Nov 2021 05:12:30 +0000 (14:12 +0900)]
ether-addr-util: make hw_addr_is_null() return true also for all zero address
This fixes the following error:
---
systemd-networkd[1449490]: ip6tnl0: Saved hardware address: 00:00:00:00:00:00:00:00:00:00:00:00:00:00:00:00
systemd-networkd[1449490]: ip6tnl0: Failed to manage link by its new hardware address, ignoring: File exists
---
userdbctl: explicitly handle ESRCH/ENOLINK from userdb_all()
Similar in style to previous commit, let's handle these two errors
properly, i.e. as equivalent to no entries found. Let's debug log about
them, to make things either to deal with when debugging (after all
userdbctl to a large degree are debugging tools).
userwork: properly handle ENOLINK error from lower-level userdb code
The lower-level userdb code will return ENOLINK if varlink lookups are
disabled explicitly and we couldn#t find an answer any other way. Let's
not propagate this to clients, since they don't have control over this
feature anyway: we decide internally when to disable varlink lookups
(e.g. if DropIn lookups are requested we disable them) and to the client
side that should not be visible: if we can't find a record with the
flags we pick then we should report then we can't find any, and that's
it.
build: preserve correct mode when generating files via jinja2
When using "capture : true" in custom_target()s the mode of the source
file is not preserved when the generated file is not installed and so
needs to be tweaked manually. Switch from output capture to creating the
target file and copy the permissions from the input file.
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
Henri Chain [Fri, 5 Nov 2021 15:11:42 +0000 (16:11 +0100)]
Use ExitType=cgroup for autostart generated services
This fixes a bug[1] with some generated autostart app services which
fork and exit immediately after main application process start,
that caused them not to launch during session startup, as the entire
cgroup was immediately killed by systemd.
This can also happen with apps such as file browsers, whose initial
process will exit after the window is closed, but who intend to leave a
daemon child running.
Since the forking model of a .desktop application cannot be known at
service generation time otherwise, ExitType=cgroup is the only effective
way to fix this bug.
Henri Chain [Wed, 24 Feb 2021 15:13:21 +0000 (16:13 +0100)]
Reintroduce ExitType
This introduces `ExitType=main|cgroup` for services.
Similar to how `Type` specifies the launch of a service, `ExitType` is
concerned with how systemd determines that a service exited.
- If set to `main` (the current behavior), the service manager will consider
the unit stopped when the main process exits.
- The `cgroup` exit type is meant for applications whose forking model is not
known ahead of time and which might not have a specific main process.
The service will stay running as long as at least one process in the cgroup
is running. This is intended for transient or automatically generated
services, such as graphical applications inside of a desktop environment.
Motivation for this is #16805. The original PR (#18782) was reverted (#20073)
after realizing that the exit status of "the last process in the cgroup" can't
reliably be known (#19385)
This version instead uses the main process exit status if there is one and just
listens to the cgroup empty event otherwise.
The advantages of a service with `ExitType=cgroup` over scopes are:
- Integrated logging / stdout redirection
- Avoids the race / synchronisation issue between launch and scope creation
- More extensive use of drop-ins and thus distro-level configuration:
by moving from scopes to services we can have drop ins that will affect
properties that can only be set during service creation,
like `OOMPolicy` and security-related properties
- It makes systemd-xdg-autostart-generator usable by fixing [1], as obviously
only services can be used in the generator, not scopes.
Dan Streetman [Fri, 5 Nov 2021 20:56:16 +0000 (16:56 -0400)]
test: refactor test-procfs-util for clarity and skip test on perm failure
After commit c3dead53d50e334f2d072a2248256983d6dc9f8c the test can fail
if the procfs file(s) aren't writable, because the check for permission
failure happens after a call that will never fail, since setting the
limit to the existing limit will always pass.
This also refactors the function slightly to make the test var names
clearer.
Peter Hutterer [Mon, 8 Nov 2021 00:47:56 +0000 (10:47 +1000)]
hwdb: remove the tablet pad entry for the UC-Logic 1060N
This entry only matches on vid/pid, so the pen event node of the device
would also get assigned the ID_INPUT_TABLET_PAD property - making it
break with libinput.
On top of that, UC-Logic's tablets re-use USB ids, so now we're breaking
multiple devices this way.
To get this device tagged correctly, use libwacom which has the
per-device hwdb entries.
projects / thirdparty / systemd.git / log
