Ok, I think I know why there is no PADO answer. The PADI is actually never going out! Now I need your help figuring out why.
I sniffed the traffic at three points:
tcpdump for eth1 on OpenWrt: I see the PADI packets with VLAN ID 7
tcpdump for eth1.7 on OpenWrt: I see the PADI packets but without VLAN header (guess this is expected?!)
Hardware TAP between router and modem, alternatively direct connection from router to laptop: There are no PPPoE packets on the wire! Only some IPv6 discovery.
So for some reason, it seems the PPPoE packets (maybe because they are VLAN tagged) do not leave the WAN interface, despite tcpdump shows them on the sender device.
What is going on there?! oO
Besides, I noticed the WAN interface goes regularly down and up again (after the PPPoE timeout). Is this normal? Example:
Thu Aug 16 08:18:55 2018 daemon.warn pppd[7460]: Timeout waiting for PADO packets
Thu Aug 16 08:18:55 2018 daemon.err pppd[7460]: Unable to complete PPPoE Discovery
Thu Aug 16 08:18:55 2018 daemon.info pppd[7460]: Exit.
Thu Aug 16 08:18:55 2018 daemon.notice netifd: Interface 'wan' is now down
Thu Aug 16 08:18:55 2018 kern.info kernel: [ 1671.594374] IPv6: ADDRCONF(NETDEV_UP): eth1.7: link is not ready
Thu Aug 16 08:18:55 2018 daemon.notice netifd: Interface 'wan' is disabled
Thu Aug 16 08:18:55 2018 kern.info kernel: [ 1671.677790] device eth1 left promiscuous mode
Thu Aug 16 08:18:55 2018 kern.info kernel: [ 1671.680162] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
Thu Aug 16 08:18:55 2018 daemon.notice netifd: Interface 'wan' is enabled
Thu Aug 16 08:18:55 2018 daemon.notice netifd: Interface 'wan' is setting up now
Thu Aug 16 08:18:55 2018 kern.info kernel: [ 1671.682434] ess_edma c080000.edma: eth1: GMAC Link is up with phy_speed=1000
Thu Aug 16 08:18:55 2018 kern.info kernel: [ 1671.687678] IPv6: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
Yes. I also was in doubt but I tried with AVM firmware and OpenWrt and my laptop with Wireshark received VLAN tagged PPPoE packets from AVM but not from OpenWrt. The latter emitted just some IPv6 discovery stuff.
So I guess either there is still something wrong with my VLAN setup (maybe because it does not appear in the switch config?!) or there is a bug somewhere.
Good idea. I already thought about how I could do this at best. Any ideas? The thing is, I guess when I send something with a packet crafting tool, it will probably be injected that low in the networking stack that it will be sent. Probably I need an application similar to pppd that I can tell to send something via eth1.7. Do you have a suitable application in mind? Otherwise I need to play around a bit..
could you further confirm that this is not due to firewall rules? (accept on default and/or wan output)
confirm this behaviour with the most default config on latest 18.06 as well as with the lede-17 branch.
then probably make bugreport along the lines vlan not working ...
I already tried latest snapshot for the previous test run. I did not test lede-17 because my device is only supported since 18.06.
I checked the firewall settings in LUCI, changed everything to accept and in the end I even flushed iptables completely. No packets go through eth1.7...
For reference here the default iptables output before I changed anything.
So I wonder whether the 4040 "routes" its WAN port via the switch? In that case you might also need to configure the switch to use VLAN tag 7 on the appropriate port?
Since @chunkeey added initial support for FRITZ!Box 4040 and @craz showed a screenshot in another thread that displays certain connectivity with this device, I'm wondering if you two have no problems with VLAN tagging on WAN port?!
I know, this thread has become a bit long, but the essence is: sending via eth1.7 does not work for me.