#20739: VLAN configuration doesn't work on the web interface (LuCi)
---------------------------+------------------------
  Reporter:  achmetow84@…  |      Owner:  developers
      Type:  defect        |     Status:  new
  Priority:  normal        |  Milestone:
 Component:  packages      |    Version:  Trunk
Resolution:                |   Keywords:
---------------------------+------------------------

Comment (by anonymous):

 I can confirm this behavior, as I ran into the exact same problem. VLANs
 created through web interface are created as eth0.XXX - but eth1  is the
 LAN interface. A manual change in the config (if you didn't lock yourself
 out of the router by this bug!) to eth1.XXX corrects the problem.

 The LUA file is the right source of the bug. Changeset 7675 from October
 2011 (http://luci.subsignal.org/trac/changeset/7675) introduced the
 heuristic to infer ethX device name from the 'swconfig dev help' command,
 with a fallback to eth0.

 At the time the lua function get_interfaces (self) line 488 is called, the
 hash baseof is initialized as an empty hash {} in line 493.

 In line 519, we are trying to lookup the base for s.device (which is
 passed into the function from the ici-lua binding - anything that looks
 like network.@switch_vlan...") which since we are looking up in an empty
 hash fails.

 Since this check fails in line 520, we attempt to match s.device against
 the regular expression "etc%d" in line 521. This fails as well, for
 instance, for me this value is 'network.@switch_vlan[0].device='switch0''

 Thus, the next call in line 523 is running swconfig dev s.device help.
 It's attempting to match that output to find ethic. Since that fails too,
 it sets base to ethic (hard coded) in line 529.

 On the Archer C7 the output of swconfig above is: switch0: ag71xx-
 mdio.0(Atheros AR8327), ports: 7 (cpu @ 0), vlans: 128
 ag71xx-mdio.0 is the BUS ID set by the network driver.


 This code probably hasn't worked on other devices such as the WDR4300, but
 incidentally, they used eth0 as the LAN interface, so the hard coded
 fallback was appropriate.

--
Ticket URL: <https://dev.openwrt.org/ticket/20739#comment:3>
OpenWrt <http://openwrt.org>
Opensource Wireless Router Technology
_______________________________________________
openwrt-tickets mailing list
openwrt-tickets@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-tickets

Reply via email to