Daan De Meyer [Wed, 26 Jun 2024 18:33:06 +0000 (20:33 +0200)]
repart: Allow Subvolumes= and DefaultSubvolume= when running offline
mkfs.btrfs has recently learned new options --subvol and --default-subvol
so let's stop failing when Subvolumes= and DefaultSubvolume= are used offline
and use the new --subvol and --default-subvol options instead to create subvolumes
in the generated root filesystem without root privileges or loop devices.
Peter Hutterer [Mon, 12 Aug 2024 04:05:13 +0000 (14:05 +1000)]
logind: warn about EVIOCREVOKE errors other than EINVAL too
EINVAL means the kernel doesn't support it, ENODEV means it's
already revoked or the device is no longer there which has the same
effect anyway. All others - let's print an error to the logs.
Yu Watanabe [Wed, 21 Aug 2024 01:23:06 +0000 (10:23 +0900)]
sd-dhcp-client: refuse to send RELEASE or friends gracefully when the daemon is stopped or so
We can easily hit the assertions without checking the internal states of
the DHCP client before calling these functions. That's annoying.
Let's do more gracefully.
Yu Watanabe [Wed, 21 Aug 2024 02:24:14 +0000 (11:24 +0900)]
sd-dhcp-client: do not call callback with SD_DHCP_CLIENT_EVENT_STOP if already stopped
When an interface enters the failed state, even if the DHCP client is
stopped, the acquired DHCP lease is not unreferenced, as the callback
dhcp4_handler() do nothing in that case. When the failed interface is
being reconfigured after that, the DHCP client is stopped again
(though it is already stopped), and SD_DHCP_CLIENT_EVENT_STOP event is
triggered and sd_dhcp_client_send_release() is called, and the
assertion in the function is triggered.
E.g.
===
systemd-networkd[98588]: wlp59s0: DHCPv4 address 192.168.86.250/24, gateway 192.168.86.1 acquired from 192.168.86.1
systemd-networkd[98588]: wlp59s0: Could not set DHCPv4 route: Nexthop has invalid gateway. Network is unreachable
systemd-networkd[98588]: wlp59s0: Failed
systemd-networkd[98588]: wlp59s0: State changed: configuring -> failed
systemd-networkd[98588]: wlp59s0: The interface entered the failed state frequently, refusing to reconfigure it automatically.
systemd-networkd[98588]: wlp59s0: DHCPv4 client: STOPPED
systemd-networkd[98588]: wlp59s0: DHCPv4 client: State changed: bound -> stopped
systemd-networkd[98588]: Got message type=method_call sender=:1.449 destination=org.freedesktop.network1 path=/org/freedesktop/network1 interface=org.freedesktop.network1.Manager member=ReconfigureLink ...
systemd-networkd[98588]: wlp59s0: State changed: failed -> initialized
systemd-networkd[98588]: wlp59s0: found matching network '/etc/systemd/network/50-wifi.network'.
systemd-networkd[98588]: wlp59s0: Configuring with /etc/systemd/network/50-wifi.network.
systemd-networkd[98588]: wlp59s0: DHCPv4 client: STOPPED
systemd-networkd[98588]: Assertion 'sd_dhcp_client_is_running(client)' failed at src/libsystemd-network/sd-dhcp-client.c:2197, function sd_dhcp_client_send_release(). Aborting.
===
Yu Watanabe [Tue, 20 Aug 2024 22:01:47 +0000 (07:01 +0900)]
network: enter initialized state when the interface will be reconfigured
When the interface is in the failed state, link_getlink_handler_internal()
will do nothing and return zero, thus the interface will not be
reconfigured, especially when the reconfiguration is triggered in
link_enter_failed().
In some kernels (specifically, 5.4) even though the clone3 syscall is
supported, setting CLONE_INTO_CGROUP is not. The error message returned
in this case is E2BIG.
If posix_spawn_wrapper encounters this error, it does not retry, and
cannot spawn any programs in said kernels.
This commit adds a check for the E2BIG error and retries pidfd_spawn()
without the POSIX_SPAWN_SETCGROUP flag.
If we encounter an E2BIG error, and the pidfd_spawn() succeeds after
removing the POSIX_SPAWN_SETCGROUP flag, then we cache the result so
that we do not retry every time.
Originally, this issue was reported in https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1077204.
Yu Watanabe [Tue, 20 Aug 2024 11:56:35 +0000 (20:56 +0900)]
test-network: do not pass '[detached]' to 'ip rule del'
That indicates the interface name in 'iif' or 'oif' cannot be resolved
when 'ip rule' command is invoked. That's natural when networkd fail to
remove rule but the corresponding interface is already removed.
To make not the residual rules interfere subsequent test cases, let's
ignore the flag and actually remove unwanted rules.
Yu Watanabe [Mon, 19 Aug 2024 22:00:44 +0000 (07:00 +0900)]
network/routing-policy-rule: also manage remaining attributes
Currently, these attributes are not configured by us, but there may be a
existing rule created by user manually with one of these attribute.
To correctly manage such foreign rules, let's read these attributes.
Yu Watanabe [Mon, 19 Aug 2024 21:19:46 +0000 (06:19 +0900)]
network/routing-policy-rule: remove rules that have conflicting flags
The kernel does not distinguish rules with different flags in
rule_exists(), but the flags of an existing rule cannot be updated.
Let's remove rules that have conflicting flags, and configure new rules
later with requested flags.
Yu Watanabe [Mon, 19 Aug 2024 19:11:19 +0000 (04:11 +0900)]
network/routing-policy-rule: anyway detach rule even when we fail to remove it
When we fail to remove a rule, that mostly means the rule does not exist
in the kernel anymore, e.g. already removed manually and we have not
received notification about that yet.
Let's detach the rule in that case.
Yu Watanabe [Mon, 19 Aug 2024 19:02:46 +0000 (04:02 +0900)]
network/routing-policy-rule: do not save rule to Manager before it is configured
Otherwise, if we fail to configure the rule, then the manager will keep
nonexistent rule forever. So, let's first copy the rule and put it on
Request, then on success generate a new copy based on the netlink
notification and store it to Manager.
Yu Watanabe [Mon, 19 Aug 2024 18:56:33 +0000 (03:56 +0900)]
network/routing-policy-rule: skip requesting when rule is already requested
If it is already requested, the new request will be anyway silently refused by
link_queue_request_safe(), which returns 0 in such case. Let's return earlier.
There should be no functional change, just refactoring.
Ronan Pigott [Mon, 19 Aug 2024 20:18:10 +0000 (13:18 -0700)]
resolved: demote the global unicast scope
This will greatly reduce the number of cases where the global unicast
scope overlaps with link scopes configured as default-route, making it
feasible to use the global DNS setting in conjunction with per-link dns
servers configured by the network.
This change is preferred over demoting links to default-route=no where
the user prefers to use the network provided DNS servers, and I expect
it is non-disruptive in that it should not degrade the efficacy of any
existing configuration.
terminal-util: don't assume errno is correctly set when using isatty_safe()
let's instead generate ENOTTY on our own. This is more correct with out
coding style (since we generally do not propagate errors via errno), and
also addresses #34039 as side effect. (#34039 really needs to be fixed
in musl though, too, this is just a work-around as side-effect).
maia x. [Mon, 19 Aug 2024 19:47:21 +0000 (12:47 -0700)]
namespace: Fix extension release memory leak
In apply_one_mount(), in the MOUNT_EXTENSION_DIRECTORY case,
char **extension_release was used as a return pointer twice but only
cleaned up once in the end. Fix it by removing duplicate code that
was causing this issue.
Yu Watanabe [Sat, 17 Aug 2024 11:13:12 +0000 (20:13 +0900)]
network/routing-policy-rule: use int32_t for suppress_prefixlen
The kernel parses FRA_SUPPRESS_PREFIXLEN as uint32_t, but internally
handled as signed integer and negative values as unset. Let's explicitly
specify the size of the variable.
Daan De Meyer [Sun, 18 Aug 2024 11:20:14 +0000 (13:20 +0200)]
test: Gracefully handle running within user namespace with single user
Unprivileged users often make themselves root by unsharing a user namespace
and then mapping their current user to root which does not require privileges.
Let's make sure our tests don't fail in such an environment by adding checks
where required to see if we're not running in a user namespace with only a
single user.
Yu Watanabe [Sat, 17 Aug 2024 02:26:32 +0000 (11:26 +0900)]
analyze: introduce --instance= option to control instance name for template units
Note, `systemd-analyze foo@.service --instance=hoge` is equivalent to
`systemd-analyze foo@hoge.service`. But, the option may be useful when
e.g. passing multiple template units that have restriction on their
instance name:
```
$ ls
template_aaa@.service template_bbb@.service template_ccc@.service
$ systemd-analyze ./template_* --instance=hoge
```
Without the option, we need to embed an instance name into each unit
name, so cannot use globs.
Before this commit, the "Cannot raise nice level" branch
is rather confusing, as we're actually lowering the nice.
Also, it's better to log about the final nice value
for both cases, no matter whether we need to set to limit
or not.
The current semantics of "filtered" in unit_is_filtered()
are actually the contrary of ListUnitsFiltered(). Let's
make things consistent, i.e. return true when the unit
shall be included.
Mike Yuan [Tue, 23 Jul 2024 14:09:53 +0000 (16:09 +0200)]
core/dbus-service: refuse bind mounting over /run/credentials/
The credential mounts should be managed singlehandedly by pid1.
Preparation for the future introduction of RefreshOnReload=credential,
where refreshing creds will be properly supported on reload.
Mike Yuan [Mon, 10 Jun 2024 15:27:51 +0000 (17:27 +0200)]
core/dbus-service: some modernization for bus_service_method_mount()
Perform some checks earlier to avoid pointless polkit auth.
Plus, the missing unit_get_exec_context() shall not be
a formalized error. As it's our internal representation
and in the normal operation should never happen.
Yu Watanabe [Fri, 16 Aug 2024 15:00:32 +0000 (00:00 +0900)]
network: make IPMasquerade= imply global IP forwarding settings again
After 3976c430927e1bfefa0413f80ebac84ab9a64350 (#31423), IPMasquerade=
implies only per-interface IP forwarding. That means, nspawn users need
to manually enable IPv4/IPv6Forwarding= in networkd.conf when
--network-veth or friend is used. Even the change was announced in NEWS,
the change itself breaks backward compatibility and extremely reduces
usability.
Let's make the setting imply the global setting again.
Daan De Meyer [Wed, 14 Aug 2024 10:43:05 +0000 (12:43 +0200)]
Add $SYSTEMD_IN_CHROOT to override chroot detection
When running unprivileged, checking /proc/1/root doesn't work because
it requires privileges. Instead, let's add an environment variable so
the process that chroot's can tell (systemd) subprocesses whether
they're running in a chroot or not.
bryango [Thu, 15 Aug 2024 05:18:17 +0000 (13:18 +0800)]
shell-completion: zsh: fix incorrect unescaping
Previously the `_filter_units_by_property` completion function
outputs with a [zsh parameter expansion flag] `g:o:`. This means
that the returned result is unescaped as the zsh builtin `echo`,
except that octal escapes don’t take a leading zero. This seemed to
have worked back in the days when it was first introduced:
udev-builtin-net_id: add NAMING_DEVICETREE_PORT_ALIASES to check of_node of netdevs before their parents
The net_id builtin only checked the of_node of a netdev's parent device,
not that of the netdev itself. While it is common that netdevs don't have
an OF node assigned themselves, as they are derived from some parent
device, this is not always the case. In particular when a single
controller provides multiple ports that can be referenced indiviually in
the Device Tree (both for aliases/MAC address assignment and phandle
references), the correct of_node will be that of the netdev itself, not
that of the parent, so it needs to be checked, too.
A new naming scheme flag NAMING_DEVICETREE_PORT_ALIASES is added to
allow selecting the new behavior.