Having the original hostname checks in the settings-service code
didn't allow the policy enough granularity to differentiate between
a plugin-provided hostname and the original hostname. We want to
fall back to the original hostname if there isn't a persistent
hostname (from a plugin) and if there isn't a DHCP-provided
hostname. Moving the original hostname checks to the policy
makes that possible. Clarify the precedence order at the same
time, and minimally validate the DHCP hostname as well.
Sometimes cause the cache not to refill with all interfaces,
meaning NM sometimes got the wrong carrier state from the
kernel which prevented NM from taking over existing connections.
nm_device_set_use_dhcp() and nm_device_get_use_dhcp() were somewhat
confusing and don't really reflect the new DHCP architecture with
NMDHCPClient. Now that timeout and state signals are specific to
the NMDHCPClient it doesn't make sense to check for DHCP use
in the callbacks for those signals since they'll never get called
if DHCP isn't in use. We might as well just keep the DHCP manager
around and check whether a DHCP client instance exists when we need
to figure out whether DHCP is in use.
Since the same interface could be used for both DHCPv4 and DHCPv6 we
can't just use 'iface' for tracking DHCP client lease changes. Instead
use a generated client ID, and track DHCP events based on the client's
PID instead of interface name.
keys- and route- files weren't passing the should_ignore_file()
check in dir_changed() because should_ignore_file() was only
taking ifcfg- files into account. Generalize most of the ifcfg name
handling functions so that should_ignore_file() will now handle
all three variants.
Add testcases to ensure that the name handling does what we want
it to, and optimize memory usage of utils_get_ifcfg_name() a bit.
