nm_exported_object_notify() hooks GObject's property-change signal
and searches for the D-Bus interface to which to send the
PropertiesChanged signal.
Then it would enqueue the value encoded as GVariant in pending_notifications.
However, thereby the association between the property that changed and the
interface was lost. So later in idle_emit_properties_changed() it would
just pick the first interface with a properties-changed-id.
That is wrong. pending_notifications must be associated with the D-Bus
interface that we are going to notify. That is, each InterfaceData must
have its own separate list.
This is broken since introducing NMExportedObject and moving to gdbus.
Only now it was discovered as NMDevice itself has two D-Bus interfaces:
"Device" and "Device.Statistics".
Note that the order of the PropertiesChanged in our D-Bus API is not defined
so that later signals can reach the receiver before earlier signals.
Also, multiple change signals for one property may be combined.
That is not changed by this patch and is not considered a bug, but something
that our D-Bus API wrt. PropertiesChanged does not guarantee.
https://bugzilla.gnome.org/show_bug.cgi?id=770629
At this point we don't know if the slave has been using an assumed
connection that just vanished -- the best bet is to let the device be.
If it's meant to be unenslaved, it won't be due to an external event.
https://bugzilla.redhat.com/show_bug.cgi?id=1357738
Now that we validate the JSON syntax of a team/team-port
configuration, any existing connection with invalid JSON configuration
would fail to load and disappear upon upgrade. Instead, modify the
setting plugins to emit a warning but still load the connection with
empty configuration.
When a software device unrealizes, we want to forget about the "user-explict"
unmanaged state. It means, that after a software device is deleted, the
"user-explict" managed flag will be cleared for that device.
It might be nice to preserve the managed-state after deletion of the device.
However, the unrealized-device only exists as long as we have a connection
for the device. That means, before this patch whether the unmanaged flag
was forgotten depends on whether the user had some connections that keep
the device alive as unrealized. That behavior was complicated, just don't
do that.
There is a "goto retry" in do_change_link_request(), at that point,
seq_result has the value -EOPNOTSUPP, instead of
WAIT_FOR_NL_RESPONSE_RESULT_UNKNOWN.
Fixes: 02fb3eff48
It seems some drivers return success for nm_platform_link_set_address(),
but at that point the address did not yet actually change *sigh*.
It changes a bit later, possibly after setting the device up.
Add a workaround to retry reading the MAC address when platform indicates
success but the address still differs at first.
https://bugzilla.gnome.org/show_bug.cgi?id=770456
Seems odd numbers may be coerced to the next-smaller even number.
Avoid that by using an even number for the test, as the number
has no particular meaning.
https://bugzilla.gnome.org/show_bug.cgi?id=765835
Depending on the connection we are about to read,
we would assert that the user provided a @out_unhandled
argument.
That means, the user must always provide a valid @out_unhandled
pointer, because he cannot know beforehand how the reading
of the ifcfg file goes.
Clear some IP related entries from the ifcfg-rh file if
the connection is a slave connection.
Also, drop utils_ignore_ip_config(). It is guaranteed, that
writer only handles connections that verify(). Such connections
have an IPv4/IPv6 setting if (and only if) they are not slave
types.
https://bugzilla.redhat.com/show_bug.cgi?id=1368761
Commit 4c7fa8dfdc ("core: drop root requirement for
load_connection(s)/set_logging D-Bus calls") removed the enforcing of
permission in the daemon for such methods since the D-Bus daemon
configuration already does that. That change also allows clients to
send a request and not wait for a response, since we don't have to
check the caller credentials in the daemon.
In the future we might switch to polkit for these methods, breaking
clients that don't wait for a reponse, so it seems better to prevent
from beginning such behavior.
Fixes: 4c7fa8dfdc
(cherry picked from commit dd27b79c4e)
The VPN data comes from an external source, it may be bogus.
Default-routes are not allowed on this point and would trigger
an assertion afterwards. Skip over them.
(cherry picked from commit 071103b172)
We need an ifindex for the NMIP4Config/NMIP6Config instance.
For interface-less VPN types, we need to lookup the parent
device, as already done for IPv4.
Fix IPv6 case too.
https://bugzilla.redhat.com/show_bug.cgi?id=1368354
(cherry picked from commit 2da35ddfe8)
When activating a connection, it may fail with nmcli reporting:
$ nmcli connection up id "Wired Connection 1"
Error: Connection activation failed: Active connection removed before it was initialized
This should be easily reproducible by having a connection "Wired Connection 1" with
cloned-mac-address set to random. When the connection is already active on a device,
re-activating with
$ nmcli connection up id "Wired Connection 1"
fails.
We first create a queued-activation and tear down the existing
connection:
device (enp0s25): state change: deactivating -> disconnected (reason 'new-activation')
Shortly after we see:
device[0x557d02cdb0c0] (enp0s25): set-hw-addr: setting MAC address to 'AA:BB:CC:DD:EE:FF' (reset, deactivate)...
device[0x557d02cdb0c0] (enp0s25): taking down device
later, we get:
device (enp0s25): link disconnected
device[0x557d02cdb0c0] (enp0s25): queued state change to unavailable due to carrier-changed (id 17290)
in the meantime, the queued activation request starts:
device (enp0s25): Activation: starting connection 'my-wired' (ca058ec5-8a47-4e1e-b38e-962b71c4699e)
but the device already transitions to unavailable
device[0x557d02cdb0c0] (enp0s25): running queued state change to unavailable (id 17290)
device (enp0s25): state change: disconnected -> unavailable (reason 'carrier-changed') [30 20 40]
which kills the new activation request:
active-connection[0x557d02c10e40]: set state deactivated (was unknown)
Just delay a carrier-lost handling if we have any queued activation
requests.
(cherry picked from commit d4e9b30320)
These logging lines are already disabled by default as _LOGt()
is a NOP unless configured --with-more-logging.
However, the logging is still very verbose also for debug-builds
and currently there are no known issues there. Disable the logging
statements (but leave them in so they can easily be enabled).
(cherry picked from commit 4cb845558e)
The D-Bus configuration already ensures that only root can do that;
enforcing the permission at policy level seems better than doing it in
the daemon itself because it allows users to change the policy and
also because callers can exit immediately after issuing the request.
(cherry picked from commit 4c7fa8dfdc)
When a assumed software device is brought down externally, it becomes
UNMANAGED_EXTERNAL_DOWN and its state goes from ACTIVATED directly to
UNMANAGED. In such case, we shouldn't flush the IP configuration
(addresses and routes) present on the device.
To fix this, clean up the device with CLEANUP_TYPE_KEEP and modify
nm_device_cleanup() not to flush addresses and devices with such flag.
https://bugzilla.redhat.com/show_bug.cgi?id=1363995
(cherry picked from commit 45cd3302dc)
Previously, we logged also the location (file:line func). nm-logging.c
supported format flags to control the timestamp, the location, and alignment
of the timestamp.
We want that all our logging backends log the same messages. That is,
both syslog and journal should have our ~default~ logging format, that
is with timestamp but without location.
Drop the unused code.
(cherry picked from commit cc828431b8)
Even if we know that the new hostname being set is equal to the cached
old one, the user may have manually changed the kernel hostname in the
meanwhile. For example:
# hostname
host123
# hostname localhost
# nmcli connection up eth1
# (now NM receives 'host123' from DHCP, but
# believes it's already set and doesn't update it)
# hostname
localhost
Let's always try to update the kernel (transient) hostname, unless it
is really already set (as returned by gethostname()).
https://bugzilla.redhat.com/show_bug.cgi?id=1356015
(cherry picked from commit 51b2cef04f)
It's not clear why a route should be suppressed if it is contained
in the subnet of one of the interface's addresses.
I think it is wrong to do this. For example, imagine an ethernet
and a Wi-Fi device both connected to the same subnet 10.0.0.0/8. By
default, ethernet gets higher priority and a better metric of 100.
If the user wants to configure a route "10.0.0.1/32 metric 99"
to reach a certain host explicitly via Wi-Fi, this check will
forbid that.
This condition was added a long time ago (38dbdae266),
but it's unclear what the original intent was.
See also commit 4f7b1cabc0, which
already relaxed this suppression of routes for non-direct routes.
(cherry picked from commit ac5dc1a951)
- don't include "nm-default.h" in header files. Every source file must
include as first header "nm-default.h", thus our headers get the
default include already implicitly.
- we don't support compiling NetworkManager itself with a C++ compiler. Remove
G_BEGIN_DECLS/G_END_DECLS from internal headers. We do however support
users of libnm to use C++, thus they stay in public headers.
(cherry picked from commit f19aff8909)
Instead of updating the device-statistic counters only periodically as
we refresh the link, update them on every link-changed event from
platform.
That means, also for devices that have RefreshRateMs at zero, the values
will be updated at random times when the link information changes.
The difference is, that previously the counters would be zero unless
RefreshRateMs was set. Now, they have some (probably stale) values
which however are not guaranteed to be kept up-to-date.
Also, now we refresh more often then promised by RefreshRateMs. But the API
technically doesn't specify that, so if we find there is a problem with
this, we may revert it later.
Originally, "nm-device-statistics.c" contained code to fetch the device
counters via netlink. As now the netlink part is handled by NMPlatform,
the code can be simplified by merging it back to NMDevice.
Technically, this is not needed because glib requires that
int is at least 32 bits. Thus, uint32 will be safely promoted
to uint.
Just do the cast to be explict about the expected type.
