When the ipv4.dhcp-fqdn property is set, NM adds the following options
to dhclient.conf:
send fqdn.fqdn "foo.bar";
send fqdn.encoded on;
send fqdn.server-update on;
which enable the S (server-update) and E (encoded) flags in DHCP
option 81, since they are sensible default values and dhclient
requires a "send fqdn.server-update [on|off]" directive in order to
send the option.
Users may want to change these flags according to their server's
configuration, but this is not possible at the moment since NM options
are placed after user's ones, overriding them.
To fix this, collect user's fqdn options and add them after NM
configuration; note that the fqdn.fqdn option still can't be
overridden by users, as NM must control the FQDN sent to server.
Fixes: c3573ebf2b
(cherry picked from commit f940428c65)
For changing the hardware address, we must bring the device down. When doing
that, IP addressing is lost and it must be re-configured after bringing the
device up again.
We already do something similar in device_link_changed(), but that might
not be sufficient, because device_link_changed() is run on an idle
handler, thus, while changing the hardware address it has no chance to
run (or notice that the device was shortly down).
https://bugzilla.redhat.com/show_bug.cgi?id=1309899
(cherry picked from commit 63571b2666)
It turnes out that 'nm_dns_manager_add_ip{4,6}_config' can be called to add VPN/DEVICE
config for device that already has one. On this case old config gets overwritten in
'priv->ip{4,6}_{vpn,device}_config' but is not removed from 'priv->configs'.
The result of that is that old config still being applyed but is treated as 'other'
config.
Apart from being a slight memory leak this has user facing consequences: if old config
was a split DNS VPN config 'split DNS' part will no longer work and those DNS servers
will be used as 'global' DNS servers.
This patch addresses the problem by making sure old values in 'priv->ip{4,6}_{vpn,device}_config'
are removed from 'priv->configs' before new values are applyed.
Signed-off-by: Nikolay Martynov <mar.kolya@gmail.com>
When the list of DNS servers changes, old DNS entries cached by
dnsmasq must be invalidated as the answers returned by new servers may
be different (especially, old NXDOMAIN entries may now be valid). Call
the dnsmasq "ClearCache" D-Bus method to achieve this.
https://bugzilla.redhat.com/show_bug.cgi?id=1338731
(cherry picked from commit 4feb58b50b)
Sometimes the netlink event lacks the IFLA_ADDRESS attribute with
the MAC address of the link. In this case, take the value from
the cached link instance. A missing netlink attribute should have the
meaning of reusing the previous value, not clearing the address.
(cherry picked from commit 791cbd0817)
To allow the execution of asynchronous actions before the system is
suspended, add a mechanism for delaying the drop of inhibitor lock.
Clients can supend the suspension by calling inhibit_take() in their
handler for SLEEPING signal and use inhibit_release() later when they
are done.
We take a "delay" type inhibitor lock, which means that the system
will proceed anyway after a certain amount of time.
Co-Authored-By: Thomas Haller <thaller@redhat.com>
(cherry picked from commit 69ab984fed)
The only user of the sleep-monitor singleton was NMManager anyway.
Also, even if we ever get more users that are interested in the SLEEPING
signal, we would hook them onto NMManager -- because NMManager should
collect, coordinate and possibly forward the SLEEPING signal. In no case,
another object should react on the SLEEPING signal and thus bypassing the
NMManager.
(cherry picked from commit 1e3b2cbb77)
Both files do very similar things, with "nm-sleep-monitor-upower.c"
being suboptimal, for example by creating the proxy synchronously.
Clean that up in the next steps. Just basic merging for now.
(cherry picked from commit c76eb3e8f7)
The lifetime of the proxy is not necesarily the same as the lifetime
of the NMSleepMonitor instance. Disconnect the signals during dispose().
(cherry picked from commit a09a5f7fc1)
The daemon does not run with a particular locale of a user. Localizing
makes no sense (at least, we don't do it usually and it would make
logging localized).
(cherry picked from commit 753f727af5)
As we don't take a reference on @self during the asynchronous
request, we must properly support cancelling in case of early
destruction.
I think, it's gdbus' responsibility not to leak any file descriptors
when cancelling a D-Bus request that returns file descriptors. Thus,
our usual pattern works here too.
(cherry picked from commit 2e3ff56cdc)
$ nmcli connection up my-connection
Connection successfully activated (D-Bus active path: /org/freedesktop/NetworkManager/ActiveConnection/42)
$ nmcli connection modify my-connection connection.id other-name
$ nmcli -f connection.id connection show other-name
connection.id: other-name
$ nmcli -f GENERAL.CONNECTION device show enp0s25
GENERAL.CONNECTION: my-connection
$ nmcli connection down other-name
Error: 'other-name' is not an active connection.
Error: no active connection provided.
$ nmcli connection down my-connection
Connection 'my-connection' successfully deactivated (D-Bus active path: /org/freedesktop/NetworkManager/ActiveConnection/13)
When modifying a connection, NMActiveConnection must update the D-Bus
properties that belong to the settings-connection.
(cherry picked from commit 18019d55a1)
This is mostly interesting of NMPolicy, which no longer needs to
subscribe to two almost identical signals (where the by-user signal
was always invoked together with the plain "updated" signal).
(cherry picked from commit 3164c026a2)
Instead of connecting to two similar signals, combine them into one
and pass "by_user" argument.
We still need to keep the original NM_SETTINGS_CONNECTION_UPDATED signal,
because it is exposed on D-Bus.
(cherry picked from commit 65c55e2a78)
I want to combine NM_SETTINGS_CONNECTION_UPDATED and NM_SETTINGS_CONNECTION_UPDATED_BY_USER
into one signal. Thus, they must have same behavior with respect as to whether they are
scheduled on idle.
Emit NM_SETTINGS_CONNECTION_UPDATED right away.
The "Updated" signal is exposed via D-Bus on the settings object.
Removing the idle handling has no bad consequences there.
Apart from that, the signal has only the listener NMSettings::connection_updated().
connection_updated() first emits NM_SETTINGS_SIGNAL_CONNECTION_UPDATED, which only
has listener: NMPolicy::connection_updated(), which already delegates to
an idle handler.
Second, NMSettings::connection_updated() emits
NM_CP_SIGNAL_CONNECTION_UPDATED which has the following listeners:
- NMBluezDevice::cp_connection_updated(), which can cause the
connection to be tracked/untracked. Any further changes are
already delayed on an idle handler (check_emit_usable_schedule()).
- NMDevice::cp_connection_added_or_updated(), which causes the
connection to be tracked by the device. It also emits
"notify::available-connections", which however is only relevant
for the D-Bus bindings.
At last, it does available_connections_check_delete_unrealized()
which also is on an idle handler already.
(cherry picked from commit a4f1a1794f)
Currently, NM_CP_SIGNAL_CONNECTION_UPDATED signal will be invoked on an
idle handler. I want to change that, so to be save, delay actions that
matter on an idle handler.
(cherry picked from commit 640ef67cd7)
schedule_activate_all() needs to iterate over all devices and is thus
relatively costly (and scales O(n^2)).
By scheduling the action on an idle handler we delay and combine
multiple redundant requests.
Another reason is that NM_SETTINGS_CONNECTION_UPDATED is currently
executed on an idle handler which first leads to
NM_SETTINGS_SIGNAL_CONNECTION_UPDATED signal and eventually calls
schedule_activate_all().
I want to change that to emit the connection update signal immediately,
thus to preserve the delay, we delay handling in NMPolicy.
(cherry picked from commit 920054d8aa)
