Log message:
Import dhcpcd-6.6.5 with the following changes:

  *  Allow STOPPED to be processed by dhcpcd-run-hooks(8)
  *  Don't attempt to match IFT to ARP types directly
  *  If neither dig nor host are present, fallback to getent(1)
  *  If IFF_UP is not set when we get a carrier up event, poll for it
     instead of giving up
  *  Failure to bind to the DHCP port on the wildcard address when
     it's in-use is not an error

Log message:
Import dhcpcd-6.6.4 with the following changes:
  *  Fix a compile warning in dhcpcd-6.6.3
  *  Notify dhcpcd-run-hooks if running in debug mode
  *  Report selected profile in test mode

Log message:
Import dhcpcd-6.6.3 with the following changes:

  *  Log which IP address we are ARPing.
  *  Only free other ARP states if not assinging an IPv4LL address.
  *  Reload global config when running per interface commands.
     This matches the signal behaviour.
  *  If we don't have a hardware address, fallback to creating a default
     IAID from the interface name and index as we used to.
  *  Loopback interfaces have routes scoped to the host only.
  *  If we don't have a hwlen and no clientid has been set, force a DUID
     based ClientID.
  *  Disable IPv6RS if a loopback, pointopoint or not a multicast interface.
  *  If allowinterfaces is not specified, allow all configured interfaces
     to work, such as loopback and ppp.
  *  When not daemonising, don't exit on timeout.
  *  Zero length UDP packets are not an error condition on the socket.
     Thanks to Micha? K?pie?.
  *  If the IP address is still on the interface when reading a lease,
     fake add the address and routes so the lease can be cleaned up if needed.

Log message:
Update dhcpcd-6.6.2 with the following changes:
  *  TAILQ macros are now pulled in via config.h only so dhcpcd compiles
     on systems where sys/queue.h does not exist at all
  *  Remove DHCP state correctly when the interface departs
  *  End the IPv4LL state when DHCP is stopped
  *  Ensure that any DHCP leased offered still exists when assigning an
     IPv4LL address

Log message:
Import dhcpcd-6.6.1 with the following changes:

  *  Log the address IPv4LL defends
  *  PREINIT, UNKNOWN and CARRIER are not either up nor down states
  *  ARP code re-written to allow for many ARP states
  *  IPv4LL address is now pseudo random based on HW address instead
     of really random as per RFC 3927 Section 2.1
  *  If not doing DHCP or DHCP6, disable the DNS requirement in the RA
     to fork.
  *  Treat IPv4LL as fallback and start DHCP discovery even if the prior
     lease was IPv4LL when rebooting.
  *  When we transition from REQUEST to DISCOVER in a reboot,
     start IPv4LL at the same time as discover to ensure we have an
     address quicker.

  *  Improve handling of the IPv6LL address at startup
  *  Support old Linux kernels where IFLA_AF_SPEC may not exist
  *  When stopping interfaces, skip past pseudo interfaces instead
     of finding the master as only the masters are sorted correctly

Added another mirror site, http://cflags.cc/roy/dhcpcd

Log message:
Import dhcpcd-6.5.1 with the following changes:

  *  Use RTF_PINNED when deleting routes when available
     Allows dhcpcd to control IPv4 routing on newer FreeBSDs
  *  Don't work on bridge, or ptp interfaces unless explicitly told
  *  Poll for IFF_RUNNING again but avoid constantly sending IFF_UP
     (should now fix all carrier problems on BSD virtual interfaces)
  *  Don't crash when processing IPv6 route calls from the kernel
     when IPv6 resources have been disabled in dhcpcd
  *  Allow the same IP address to be shared across different interfaces
     Interface with the lowest metric gets the IP address, will move
     to the next highest if dropped (interface departs, carrier drops, etc)
  *  Use correct interface gateway on FreeBSD, removes need for linkaddr.c
     on kFreeBSD
  *  Delegated prefix addresses are now reported via DELEGATE6
  *  Fix copying the correct timezone file
  *  Work better with unknown delegated prefix lengths
  *  Move IPv4LL and ARP to the DHCP eloop queue to fix timing issues
  *  Add IA PD documentation update from christos@netbsd.org

Log message:
Import dhcpcd-6.5.0 with the following changes:

  *  Fix an unaligned access error on BeagleBone Black with FreeBSD.
     Thanks to Guy Yur for the patch.
  *  Remove the fast loop trying to up an interface which does not
     report carrier.
  *  Remove vis based encoding - instead validate against option type and
     stop at invalid [1]
     This removes all shell escaped encoding - dhcpcd will assume that IF
     the --script option is a shell, it will quote variables correctly.
     The stock dhcpcd-run-hooks does.
  *  dhcpcd -V now prints how the variables will be decoded.
  *  Changed some options in dhcpcd-definitions.conf to more sensible defaults.
  *  Don't daemonise on delegated address dad.
  *  Don't drop delegated reject route when forking.
  *  Fix IPv6 handling of link-local addresses on KAME stacks.
  *  Work on OpenBSD-5.6 without any special interface setup needed.
  *  Callout to handlecarrier when we don't have real carrier support and
     rely on looking at IFF_UP and IFF_RUNNING.
     This allows our hooks to know that dhcpcd thinks we have a carrier or not.

[1] DHCP option encodings defined in dhcpcd-definitions.conf
  *  domain (RFC3397)/dname (string) is strict domain name allowance
     (ie, [alnum] with _- (but not at the start or end))
  *  string is now printable ascii (1-127) until invalid
  *  ascii is all ascii (1-127) until invalid
  *  raw is all chars (1-255) until NUL
  *  binhex is a hex representation of the option including embedded NULs
  *  ssid is still escpaed octal because it's expected to be human readable
     AND can technically be all NUL
  *  everything else has strict option -> value encoding

Log message:
Roll back to dhcpcd-6.4.3

Log message:
Import dhcpcd-6.4.7 with the following changes:
  *  make test works again
  *  Many bounds checking fixes from Tobias Stoeckmann
  *  Improve error when the authentication token cannot be found
  *  close the IPv4 specific UDP socket when done sending
  *  Implemented a write queue to the control sockets
  *  Only send interfaces to control sockets when in a BOUND state
  *  Add a sample controlgroup directive to dhcpcd.conf to make setup easier
  *  Add variables if_oneup and if_ipwaited so hook scripts know the overall
     state of dhcpcd better
  *  Pass RC_SVCNAME from enviromment to hooks so that a service hook can
     know it's name (may not be dhcpcd)
  *  Document every variable set for dhcpcd-run-hooks(8)
  *  Use the nl80211 interface on Linux to get the wireless SSID if we fail
     to get it via WEXT
  *  Allow SSIDs with non printable characters to be used in ssid selection
     in dhcpcd.conf
  *  Add an unprivileged control socket so that normal users can obtain
     dhcpcd running state
  *  Remove all instances of if_indextoname as we already know the index
  *  Only bring in linux/ipv6.h for linux AND glibc
  *  Add _DEFAULT_SOURCE #define to to make glibc-2.20 happy
  *  Check we have allocated IPv6 resources before checkings RA's
  *  configure errors are now logged to config.log
  *  Only hunt for a cross compiler if build != host
  *  Detect removal of IPv6 routes
  *  Don't add link-local addresses to POINTOPOINT interfaces
  *  Don't discard expired DHCPv6 leases when dumping them
  *  If a DHCPv6 lease has no timers, expire it right away
  *  Report delegated addresses
  *  Call dhcpcd-run-hooks correctly when delegated prefixes already exist
  *  Fix a memory error when ia_* config exists but IPv6 is disabled
  *  Ensure servername and bootfile are safely exported
  *  Sanitise the following characters using svis(3) with VIS_CTYLE and
     VIS_OCTAL:
         | ^ & ; < > ( ) $ ` \ " ' <tab> <newline>
     This allows a non buggy unvis(1) to decode it 100% and stays compatible
     with how dhcpcd used to handle encoding on most platforms.
     For systems that supply svis(3) there is a code reduction, for systems
     that do not, a slight code increase. This change mitigates systems
     affected by bash CVE-2014-6271 and CVE-2014-7169.

OK: jperkin@

Log message:
Update to dhcpcd-6.4.3 with the following changes:

  *  Correct DHCPv6 Prefix Delegation option decoding
  *  Ensure that a given buffer is at least BUFSIZ (for getline compat)
  *  Poll interfaces on BSD for IFF_RUNNING if link state cannot be obtained
  *  Check for an IA to use in DHCPv6 lease validation
  *  Fix compile on NetBSD-6 (and possibly earlier NetBSDs)
  *  Warn about exceeding IDGEN_RETRIES when a stable private address cannot
     be obtained
  *  Fix DHCP option overload handling, thanks to Tobias Stoeckmann