Re: CFT: cloned_interfaces and gifconfig in rc.d/netif

Hiroki Sato <hrs_at_freebsd.org> wrote
  in <20130722.024513.95685108976349294.hrs_at_allbsd.org>:

hr> Hi,
hr>
hr>  The attached patch allows rc.d/netif to create IFs in
hr>  $cloned_interfaces when interface name(s) is specified.  For example,
hr>  when the following lines are in rc.conf:
hr>
hr>   cloned_interfaces="bridge0 bridge1"
hr>   ifconfig_bridge0="..."
hr>   ifconfig_bridge1="..."
hr>
hr>  The following commands create the interfaces and destroy them.
hr>
hr>   # /etc/rc.d/netif start bridge0 bridge1
hr>   # /etc/rc.d/netif stop bridge0 bridge1
hr>
hr>  netif cloneup/clonedown does this without the patch, but it cannot
hr>  configure the interfaces and does not support clean teardown.
hr>
hr>  Also, routines which handle $gif_interfaces are merged into ones for
hr>  $cloned_interfaces.  ifconfig_gifN and other variants did not work
hr>  with gif interfaces defined in $gif_interfaces.  The patch solves
hr>  this issue.
hr>
hr>  Basically there should be no functionality regression for the
hr>  existing configurations.  Can anyone who are using $gif_interfaces
hr>  and/or $cloned_interfaces test this?  I would like to know if there is
hr>  regression or not.

 A revised version based on feedback which I received in private
 emails is attached.  This includes rc.conf(5) change explaining the
 new variables.  The following is a brief description of the changes:

-----
- Reimplement gif_interfaces as a variant of $cloned_interfaces.
  Newly-configured systems should use $cloned_interfaces.

- Call clone_{up,down}() in rc.d/netif {start,stop}.

- Add rc.d/netif clear.  The "clear" argument is basically equivalent to
  "stop" but it does not call clone_down().

- Add "ifname:sticky" keyword into $cloned_interfaces.  If :sticky is
  specified, the interface will not be destroyed in rc.d/netif stop.

- Add cloned_interfaces_sticky={YES,NO}.  This variable globally sets
  :sticky keyword above for all interfaces.  The default value is NO.
----

-- Hiroki