BIND 10 #1238: IfaceMgr support for IPv4 (incl. Socket binding for v4) (was: Socket binding for v4)

[BIND 10 Development]

#1238: IfaceMgr support for IPv4 (incl. Socket binding for v4)
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  UnAssigned
  stephen                            |                Status:  reviewing
                       Type:  task   |             Milestone:  Sprint-
                   Priority:  major  |  DHCP-20111123
                  Component:  dhcp   |            Resolution:
                   Keywords:         |             Sensitive:  0
            Defect Severity:  N/A    |           Sub-Project:  DHCP
Feature Depending on Ticket:         |  Estimated Difficulty:  0
        Add Hours to Ticket:  0      |           Total Hours:  0
                  Internal?:  0      |
-------------------------------------+-------------------------------------
Changes (by tomek):

 * owner:  tomek => UnAssigned
 * status:  accepted => reviewing


Comment:

 As IfaceMgr will be shared between DHCPv4 and DHCPv6, it has to be moved
 to src/lib/dhcp eventually. I believe it is too early for that step, but
 won't object is reviewer suggest otherwise. On a practical aspect, the
 most convenient moment to move it is once DHCPv4 daemon is implemented.
 That will happen after send and receive for IPv6 is implemented.

 As this ticket introduced support for IPv4 sockets to IfaceMgr, number of
 extra features not strictly related to socket binding for IPv4 were
 implemented. Therefore renamed this ticket.

 Added generic mechanism for associating sockets with interfaces. That is
 an aspect in DHCP 4 that caused problems on several occasions. This
 approach also allows very simple implementation of listening over several
 interfaces.

 As part of generic IfaceMgr improvement efforts, some previously disabled
 tests are now enabled. One multicast-oriented test has to remain disabled,
 as it does not work on NetBSD. That system is currently out of scope, and
 it is not reasonable to make it platform-agnostic before interface
 detection is implemented for all systems. This test uses link-scoped
 multicast addresses.

 I was considering adding family enum, but eventually chose to use AF_INET
 and AF_INET6 to remain consistent with values returned by
 asiolink::IOAddress::getFamily().

 Code is implemented and ready for review.

-- 
Ticket URL: <http://bind10.isc.org/ticket/1238#comment:3>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development