Next we will need to detect more kernel features. First refactor the
handling of these to require less code changes and be more efficient.
A plain nm_platform_kernel_support_get() only reqiures to access an
array in the common case.
The other important change is that the function no longer requires a
NMPlatform instance. This allows us to check kernel support from
anywhere. The only thing is that we require kernel support to be
initialized before calling this function. That means, an NMPlatform
instance must have detected support before.
(cherry picked from commit ee269b318e)
Currently, if user configuration or settings specify that a software
device is unmanaged, for example:
[device-bond-unmanaged]
match-device=interface-name:bond*
managed=0
or
[keyfile]
unmanaged-devices=interface-name:bond*
and there is a connection for the device with autoconnect=yes, NM
creates the platform link and a realized device in unmanaged
state. Fix this, the device should not be realized if it is unmanaged.
https://bugzilla.redhat.com/show_bug.cgi?id=1679230
nm_device_spec_match_list_full() calls
nm_device_get_permanent_hw_address() which freezes the MAC address, so
currently callers must avoid the function when the device is not
completely platform-initialized.
Instead, use nm_device_get_permanent_hw_address_full() to avoid
freezing the MAC when the device is not platform-initialized. In this
way nm_device_spec_match_list_full() can be called from any state
without side effects.
In general shortcutting state is a no-no. But putting a device to FAILED
state because its master is going down is a crime. It's the wrong state:
the devices should enter it when their connections themselves failed
unexpectedly, and can potentially recover with another actiation.
Otherwise bad things happen,
In particular, the devices automatically enter DISCONNECTED state and
eventually retry autoconnecting. In this case they would attempt to
bring the master back up. Ugh.
This situation happens when a topomost master of multiple levels of
master-slave relationship is deactivated.
Aside from that, shortcutting to DISCONNECTED on unknown change reason
doesn't make sense either. Like, wtf, just traverse through DEACTIVATING
like all the other kids do.
Connection defaults should correspond in range to the per-profile values.
"infiniband.mtu" is required to be not larger than 65520, so we also
need to honor that when parsing the connection default.
... and nm_acd_manager_announce_addresses().
The test will need more information to know why it may fail.
Return a NetworkManager error code, instead of a boolean.
If we surprise-remove the master, slaves would immediately attempt to bring
things up by autoconnecting. Not cool. Policy, however, blocks
autoconnect if the slaves disconnect due to "dependency-failed", and it
indeed seems to be an appropriate reason here:
$ nmcli c add type bridge
$ nmcli c add type dummy ifname dummy0 master bridge autoconnect yes
$ nmcli c del bridge
$
Before:
(nm-bridge): state change: ip-config -> deactivating (reason 'connection-removed')
(nm-bridge): state change: deactivating -> disconnected (reason 'connection-removed')
(nm-bridge): detached bridge port dummy0
(dummy0): state change: activated -> disconnected (reason 'connection-removed')
(nm-bridge): state change: disconnected -> unmanaged (reason 'user-requested')
(dummy0): state change: disconnected -> unmanaged (reason 'user-requested')
policy: auto-activating connection 'bridge-slave-dummy0'
After:
(nm-bridge): state change: ip-config -> deactivating (reason 'connection-removed')
(nm-bridge): state change: deactivating -> disconnected (reason 'connection-removed')
(nm-bridge): detached bridge port dummy0
(dummy0): state change: activated -> deactivating (reason 'dependency-failed')
(nm-bridge): state change: disconnected -> unmanaged (reason 'user-requested')
(dummy0): state change: deactivating -> disconnected (reason 'dependency-failed')
(dummy0): state change: disconnected -> unmanaged (reason 'user-requested')
https://github.com/NetworkManager/NetworkManager/pull/319
When the link goes down the kernel removes IPv6 addresses from the
interface. In update_ext_ip_config() we detect that addresses were
removed externally and drop them from various internal
configurations. Don't do that if the link is down so that those
addresses will be restored again on link up.
(cherry picked from commit 505d2adbc2)
Add a new argument to nm_ip_config_* helpers to also ignore addresses
similarly to what we already do for routes. This will be used in the
next commit; no change in behavior here.
(cherry picked from commit 39b7257208)
We can detect false DAD failures if the link goes down. Don't try to
prevent them, but just reset the counter if the link goes down.
(cherry picked from commit 056470a4ba)
When the interface is down DAD failures becomes irrelevant and we
shouldn't try to add a link-local address even if the configuration
contains other IPv6 addresses.
(cherry picked from commit 72385f363c)
The device type was set to the GType rather than a new value in the
NMDeviceType enum.
Add the corresponding enum entry, fix the device type and set the
routing priority to the same value as generic devices.
(cherry picked from commit 8d9365a973)
dev2_ip_config (formerly wwan_ip_config) is only set by nm_device_set_dev2_ip_config()
(formerly nm_device_set_wwan_ip_config()), which is only called by NMDeviceModem.
For NMDeviceWireGuard we will also inject additional configuration
in the parent class. Rename and give it a wider purpose. The new name
merely indicates that this IP configuration is injected by a subclass
of NMDevice.
(cherry picked from commit 03b708f7f7)
Now they follow the naming pattern of ending in "_4" / "_6".
We will merge them and alias them to an "_x" array, like done
for similar fields.
(cherry picked from commit ca14df5619)
It is preferable to treat IPv4 and IPv6 in a similar manner.
This moves the places where we differ down the call-stack.
It also make it clearer how IPv6 behaves differently. I think this
is a bug, but leave it for now.
+ /* If IP had previously failed, move it back to IP_CONF since we
+ * clearly now have configuration.
+ */
+ if (priv->ip6_state == IP_FAIL)
+ _set_ip_state (self, AF_INET6, IP_CONF);
(cherry picked from commit 1585eaf473)
Wrongly did not suppress the message
<warn> [1550844832.3749] device (tunl0): failed to disable userspace IPv6LL address handling (not-supported)
Fixes: d18f40320d
Delay ARP announcements for masters until the first interfaces gets
enslaved. There is no point in doing it before as the ARP packets
would be dropped in most cases; also, if the first slave is added when
we already started announcing, the MAC of the master is going to
change and so the remaining ARPs will have a wrong "sender mac
address" field.
https://bugzilla.redhat.com/show_bug.cgi?id=1678796https://github.com/NetworkManager/NetworkManager/pull/301
Usually, for external/assume we skip calling act_stage2_config().
Add a flag that allows the device to indicate that it always wants
to be called. This is useful, if the device wants to do some initialization
also for external/assume cases.
Instead of performing a series of steps inside one check for
"!nm_device_sys_iface_state_is_external_or_assume (self)", perform
all steps individually (under the same check).
There is no change in behavior, but this is more logical to me.
We perform a series of steps, depending on condition. Each step
individually depends on a set of conditions, instead of checking
for a set of conditions and doing a series of independent steps.
WireGuard devices are (will be) regular NMDevice implementations,
but NMDnsManager should treat them like VPN.
For that, reuse the device's type and nm_device_get_route_metric_default().
For static functions inside a module, the compiler determines on its own
whether to inline the function.
Also, "inline" was used at some places that don't immediatly look like
candidates for inlining. It was most likely a copy&paste error.
NM_UTILS_LOOKUP_STR() uses alloca(). Partly to avoid the overhead of
malloc(), but more important because it's convenient to use. It does
not require to declare a varible to manage the lifetime of the heap
allocation.
It's quite safe, because the stack allocation is of a fixed size of only
a few bytes. Overall, I think the convenience that we get (resulting in
simpler code) outweighs the danger of stack allocation in this case. It's
still worth it.
However, as it uses alloca(), it still must not be used inside a (unbound)
loop and it is obviously a macro.
Rename the macros to have a _A() suffix. This should make the
peculiarities more apparent.
