As with the other connection-matching methods, move the loop and the
device-independent bits into NMDevice. By reusing
nm_device_check_connection_compatible(), this means that most device
types now no longer need any type-specific code for this.
https://bugzilla.gnome.org/show_bug.cgi?id=693684
nm_device_connection_match_config() sounded more generic than it
really was; rename it to nm_device_find_assumable_connection(), which
is what it really does.
There was also a lot of redundancy/cut+paste in the subclass
implementations of connection_match_config(); Improve things by moving
the looping-over-connections code into NMDevice itself, and also doing
the general-device-compatibility and IP-config checking there, leaving
the device subclasses to just verify L2 properties. Which most of them
aren't doing...
https://bugzilla.gnome.org/show_bug.cgi?id=693684
Since NMDevice has a generic get_hw_address() method now, it can do
nm_device_spec_match_list() itself (for everything except ethernet,
which needs to match against s390 subchannels too).
https://bugzilla.gnome.org/show_bug.cgi?id=693684
in hw_take_down()
usb 1-3: USB disconnect, device number 6
modem-manager[547]: <info> (tty/ttyACM0): released by modem /sys/devices/pci0000:00/0000:00:1a.7/usb1/1-3
<info> (tty/ttyACM0): released by modem /sys/devices/pci0000:00/0000:00:1a.7/usb1/1-3
<info> (ttyACM0): device state change: disconnected -> unmanaged (reason 'removed') [30 10 36]
<info> (ttyACM0): cleaning up...
<info> (ttyACM0): taking down device.
nm_system_iface_set_up: assertion `ifindex > 0' failed
==23089== 38,232 (4,248 direct, 33,984 indirect) bytes in 177 blocks are definitely lost in loss record 5,122 of 5,123
==23089== at 0x4A0881C: malloc (vg_replace_malloc.c:270)
==23089== by 0x39B905488E: g_malloc (gmem.c:159)
==23089== by 0x39B9068CA1: g_slice_alloc (gslice.c:1003)
==23089== by 0x39B98371B1: g_value_array_new (gvaluearray.c:140)
==23089== by 0x31FB81B67D: valuearray_constructor (dbus-gvalue-utils.c:771)
==23089== by 0x42DD8F: get_property (nm-device.c:4675)
==23089== by 0x39B9819C64: g_object_get_property (gobject.c:1289)
==23089== by 0x31FB80DA49: object_registration_message (dbus-gobject.c:1322)
==23089== by 0x363961DA44: _dbus_object_tree_dispatch_and_unlock (dbus-object-tree.c:858)
==23089== by 0x363960FA82: dbus_connection_dispatch (dbus-connection.c:4685)
==23089== by 0x31FB80AC44: message_queue_dispatch (dbus-gmain.c:90)
==23089== by 0x39B904EC54: g_main_context_dispatch (gmain.c:2539)
Otherwise, priv->accept_ra_path would be NULL, which isn't very
useful and makes nm_utils_do_sysctl() angry. No reason we shouldn't
always create priv->accept_ra_path in the future though.
https://bugzilla.gnome.org/show_bug.cgi?id=691213
Add a "need_carrier" argument to nm_device_is_available(), to allow
distinguishing between "device is not available", "device is fully
available", and "device is available except for not having carrier".
Adjust various parts of NMDevice and NMManager to allow for the
possibility of activating a connection with :carrier-detect = "no" on
a device with no carrier, and to avoid auto-disconnecting devices with
:carrier-detect = "on-activate".
https://bugzilla.gnome.org/show_bug.cgi?id=688284
Move some duplicated carrier-handling code into NMDevice (which can
introspect itself to see if it's a subclass that has carrier).
The "mostly ignore carrier" special handling for bridges and bonds is
now also handled as part of the NMDevice-level carrier handling.
https://bugzilla.gnome.org/show_bug.cgi?id=688284
g_malloc(), etc, never return NULL, by API contract. Likewise, by
extension, no other glib function ever returns NULL due to lack of
memory. So remove lots of unnecessary checks (the vast majority of
which would have immediately crashed had they ever run anyway, since
g_set_error(), g_warning(), and nm_log_*() all need to allocate
memory).
https://bugzilla.gnome.org/show_bug.cgi?id=693678
The hardware address of a device often changes when slaves change,
for example with bond interfaces, which take the hardware address of
their first slave. Make sure that's always up-to-date.
It's not very helpful to deactivate a slave of a bond or a bridge
port because the carrier is lost. Carrier failures are normal
operation when using various bonding failover modes. For bridge
ports, it's probably not very helpful either. Given that the
masters are all virtual interfaces and have been explicitly
configured and started by a user, the configuration should
generally be left alone unless explicitly deactivated or changed.
Two things:
1) When the slave was deactivated, nm_device_deactivate() runs before
the master gets the slave's state-changed signal, and thus priv->master
is cleared long before nm_device_notify_enslaved() is called. Which
would trigger the g_assert (priv->master).
2) If the slave is already deactivated, there's no point in re-queueing
a state change to deactivated.
So just assert that priv->master is valid if the slave is going to be
enslaved, but if the slave is being released, ignore NULL priv->master
which we don't use anyway. Also ignore redundant state changes.
Instead of using a mix of netlink and SIOCGIFHWADDR and doing it
in every device, create a utility function for this and have
everywhere else call that.
We'll want to use NMActiveConnection more in the manager and also fold
the PendingActivation functionality into it. All this functionality
applies to the VPN connections too, so it makes sense to have it all
in the base class instead of both NMActRequest and NMVPNConnection.
If, for example, a bond interface has dynamic IPv4 configuration
and static IPv6 configuration, then without slaves IPv6 config
can proceed but IPv4 cannot until a slave is present. Allow
subclasses to postpone a specific IP configuration path until
they're ready, but let others proceed.
We don't need to check device state here because the manager, which
is the only thing that calls nm_device_activate() in
internal_activate_device() ensures that the device is deactivated
before starting a new activation request.
This function gets used for both /proc/sys (ie, sysctl) and for
sysfs attributes. There are two issues with it:
1) most sysctl values don't care about a trailing LF, but some
sysfs attributes (infiniband) do; so we always have to add the
trailing LF. Just move that into the function to ensure that
callers don't forget to add it.
2) neither sysfs or sysctl support partial writes, while the
existing function did partial writes. Practically, both the
write handlers for sysfs and sysctl should always handle all
the data, but if they don't, partial writes are wrong. So
instead, try three times to write all the data.
DHCP lease file names are built by NetworkManager and contain connection UUID
which makes them NM-specific. Their new location belongs to NetworkManager and
doesn't have to be guessed. With no guessing, we don't need distribution-specific
conditionals.
Note: This change may require modifications to the selinux policy. But after all
these files actually belong to NetworkManager as well as the instance of dhclient
that uses them.
Rather than having NMDevice subclasses connect to their own
::state-changed signal, fix up the signal definition so they can just
override the class handler.
The idea was copied from gtk, but it's only used there in cases where
the method's wrapper function and default implementation would
otherwise have the same name, which never happens in NM because our
method implementations aren't prefixed with the type name, so it's
just noise here.
VLANs are only supported on certain kinds of devices, so don't try to
create them on other devices. (In fact, NM currently assumes that
VLANs are only created on Ethernet devices, so we need to be even more
picky than that.)
