Daan De Meyer [Tue, 28 Mar 2023 10:32:51 +0000 (12:32 +0200)]
btrfs-util: Add BTRFS_SNAPSHOT_LOCK_BSD
When making ephemeral snapshots of subvolumes whose cleanup depends on
whether they're locked or not, it's necessary to have the lock from the
very beginning, so let's support that with a new BTRFS_SNAPSHOT_LOCK_BSD
flag.
Daan De Meyer [Fri, 24 Mar 2023 16:12:24 +0000 (17:12 +0100)]
copy: Add COPY_LOCK_BSD
When making ephemeral copies of files/directories whose cleanup
depends on whether they're locked or not, it's necessary to have the
lock from the very beginning, so let's support that with a new
COPY_LOCK_BSD flag.
This adds two things: a note to the --help text that people use the
relevant systemctl commands instead (as they are a lot more powerful,
for example give you inhibitor and boot loader control, kexec, and so
on). And a note to developers that they stop adding new stuff to the
compat interfaces.
Yu Watanabe [Thu, 15 Jun 2023 05:00:09 +0000 (14:00 +0900)]
kernel-install: skip to read /etc/machine-info in test
We do not provide any way to override /etc/machine-info.
As the file is deprecated in kernel-install, let's skip to read it when
we test kernel-install.
Henrik Holst [Thu, 15 Jun 2023 04:32:10 +0000 (06:32 +0200)]
network: make degraded-carrier bond/bridge as routable (#27776)
This makes a bond or bridge interface in the degraded-carrier state but has a routable address
handled as routable operational state.
If the carrier is degraded but the address state is routable then the operational state should be
seen as routable and not degraded because that may be the case for bonds if some of the links are down,
but when that happens the bond as whole is still routable.
This also makes operational state to degraded if address state is degraded even if the link state is
degraded-carrier.
Frantisek Sumsal [Wed, 14 Jun 2023 18:14:25 +0000 (20:14 +0200)]
test: exit on first failing subtest
Let's take a step back and revert back to the original behavior where we
exit on a first failing subtest. The current behavior makes fishing out the
failing test details quite unpleasant, and in certain situations the
journal may even be rotated away so we end up with no actionable logs.
Daan De Meyer [Wed, 14 Jun 2023 08:27:22 +0000 (10:27 +0200)]
mkfs-util: Hide /proc/self/mounts before running mkfs
mkfs.btrfs refuses to operate on a block device with mounted
partitions, even if doing so is perfectly safe. An example when
this happens is when using systemd-repart with it's --image switch
to add a root partition to a /usr only image. As a workaround until
the issue is fixed, let's hide the information on mounted filesystems
from mkfs.btrfs so it doesn't fail and formats the new filesystem as
expected.
bootctl: warn if the ESP random seed is stored on a world-readable dir
This takes heavy inspiration from @zx2c4 (Jason A. Donenfeld)'s
PR #25531 but changes it considerably, but always going by fd instead of
paths, and only warning about the side file itself and the ESP mount
point, nothing else. This shuld be more than enough and should not be
brittle against concurrent path modifications.
The idea is to make it easy to generate all the signing key and certs
that can be used for local signing. The verb is the modeled after
'mkosi genkey', but there are some important differences: we generate
the keys to the paths where they will be read from, both pcr signing
keys and the SecureBoot certificate+key.
If any of the outputs exist, operation is refused. Maybe we could add a
--force option in the future, but this operation should be rare, so I think
it's better to refuse to overwrite anything initially.
I'm only doing a token man page change here.
https://github.com/systemd/systemd/pull/27621 reworks the man page,
and the changes done here would conflict heavily with that work. I'll
submit a follow-up patch later.
tree-wide: when in doubt use greek small letter mu rather than micro symbol
Doesn't really matter since the two unicode symbols are supposedly
equivalent, but let's better follow the unicode recommendations to
prefer greek small letter mu, as per:
time-util,socket: accept both kinds of unicode µ symbols
Apparently there are two µ symbols, accept both when parsing.
One is the greek small letter mu (μ) the other is the micro sign (µ).
Unicode recommendation considers both equivalent, and says use of greek
small letter mu is preferred. See:
Yu Watanabe [Fri, 13 Jan 2023 04:25:43 +0000 (13:25 +0900)]
udev-node: optimize device node symlink creation
If multiple devices requested the same device node symlink with the same
priority, then previously we read O(N^2) of files saved in
/run/udev/links.
This makes if the requested symlink already exists with equal or higher
priority, then the symlink is kept, and skip to read all existing files,
except for one related to the current device node, in /run/udev/links.
Hence, the total amount of file read becomes O(N).
This improves performance of testcase_simultaneous_events_2 added by the
previous commit about 30%.
Before (32.8 sec):
```
## 3 iterations start: 11:13:44.690953163
## 3 iterations end: 11:14:17.493974927
```
After (23.8 sec):
```
## 3 iterations start: 11:17:53.869938387
## 3 iterations end: 11:18:17.624268345
```
This is based on the idea and analysis by Franck Bui.
Daan De Meyer [Tue, 13 Jun 2023 13:52:03 +0000 (15:52 +0200)]
repart: Store dm_name in DecryptedPartitionTarget
This means we don't have to do a fallible allocation in the
DecryptedPartitionTarget destructor. Also use log_warning_error_errno()
for the failure we ignore in the destructor.
Daan De Meyer [Thu, 8 Jun 2023 15:50:05 +0000 (17:50 +0200)]
mkosi: Remove explicit /testok check
vsock should work properly after the latest release of mkosi. But
to make sure it works, let's exit with 123 in case of success and
check for that in Github Actions.
Daan De Meyer [Tue, 13 Jun 2023 08:26:00 +0000 (10:26 +0200)]
sd-daemon: Add vsock fallback to SOCK_STREAM
SOCK_SEQPACKET is only supported on recent kernels. Let's add a
fallback to SOCK_STREAM if it is not supported.
To accomodate SOCK_STREAM, we also modify
pid_notify_with_fds_internal() to do sendmsg() in a loop as with
SOCK_STREAM, sendmsg() is not guaranteed to write all data in a single
syscall or fail otherwise.
test: merge TEST-61-UNITTESTS-QEMU with TEST-02-UNITTESTS
The test was originally introduced for Ubuntu CI, as it uses
PREFER_NSPAWN=1, but it was subsequently disabled two years ago [0], so
it was pretty much useless. Let's merge it into TEST-02 and tweak it a
bit to run only certain tests under QEMU when $TEST_PREFER_NSPAWN is
set.
Frantisek Sumsal [Mon, 12 Jun 2023 19:05:30 +0000 (21:05 +0200)]
proc-cmdline: parse the whole /proc/cmdline
The kernel command line may contain newlines which kernel happily
accepts, but we'd ignore everything past the first newline. Let's fix
that by replacing read_one_line_file() with read_full_file().
tmpfile-util: turn last parameter of link_tmpfile() into a proper flags
This changes a boolean param into a proper bitflag field.
Given this only defines a single flag for now this doesn't look like
much of an improvement. But we'll add another flag shortly, where it
starts to make more sense.
resolved: add DumpCache varlink call for acquiring a complete dump of all of resolved's RR caches
This adds a simple varlink call io.systemd.Resolve.Monitor.DumpCache to
the existing io.systemd.Resolve.Monitor service. It compiles a JSON
object containing the per-scope cache entries and returns it.
Jan Janssen [Mon, 12 Jun 2023 13:12:05 +0000 (15:12 +0200)]
boot: Unify protocol opening
We are using HandleProtocol everywhere except in these few cases. This
unifies on HandleProtocol as it is simpler to use and equivalent to
OpenProtocol.
The only difference between the two is that OpenProtocol attributes the
opened protocol to a firmware-owned handle instead of our image handle.
This has no real use for regular UEFI applications as any protocols
opened via BY_HANDLE or GET_PROTOCOL is not required to be closed. In
fact, when a protocol is uninstalled it will do nothing more than reduce
the open count for these.
This is identical to json_dispatch_variant() but doesn't increase the
ref counter of the variant. This is useful for taking JSON objects apart
locally without the intention to keep any component pinned for longer
than the local stack frame.
This simplifies the logic: finalize_options() is the step that does the
checks and the mangling. The checks for consistency are done in more places,
so we need to pass a verb (we only have 'build', but once we add other verbs,
any would do).
This changes the generated config.h file thusly:
-#define _GNU_SOURCE
+#define _GNU_SOURCE 1
Canonically, _GNU_SOURCE is just defined, without any value, but g++ defines
_GNU_SOURCE implicitly [1]. This causes a warning about a redefinition during
complilation of C++ programs after '-include config.h'. Our config attempts to
inject this (and a bunch of other arguments) into all compliations. But before
meson 0.54, flags for dependencies were not propagated correctly (*), and the C++
compilation was done without various flags (**). Once that was fixed, we started
getting a warning.
[1] http://gcc.gnu.org/onlinedocs/libstdc++/faq.html#faq.predefined
(*) Actually, the changelog doesn't say anything. But it mentions various work
related to dependency propagation, and apparently this changes as a side
effect.
(**) -fno-strict-aliasing
-fstrict-flex-arrays=1
-fvisibility=hidden
-fno-omit-frame-pointer
-include config.h
This could be solved in various ways, but it'd require either making the
compilation command line longer, which we want to avoid for readability of the
build logs, or splitting the logic to define the args for C++ progs separately,
which would make our meson.build files more complicated. Changing the
definition to '1' also solves the issue (because apparently now we match the
implicit definition), and shouldn't have other effects. I checked compilation
with gcc and clang. Maybe on other systems this could cause problems. We can
revisit if people report issues.
projects / thirdparty / systemd.git / log
