On 11/12/2015 14:48, Vittorio G (VittGam) wrote: > Hi, > > On 11/12/2015 14:32:45 CET, John Crispin wrote: >> hardcoding the board id here is not a good idea. we should rather use >> the board.json data to probe if a switch is registered. if not we dont >> apply the hack. i'll ask jow to tell us how to probe for a switch. > > We cannot use board.json data here: this code is called before mounting > the rootfs, in the preinit.
we can. i just spoke with jow and we will add a codepath to failsafe to probe board.json > > Also, the switch is actually registered, just vlans are not needed on > HT-TM02 for normal operation and thus are normally disabled. we will be able to derive the correct setup directly from board.json in that case and wont have to reply on this workaround that we have had for what feels like far too long > > Also, the LAN and only port is 4, so by adding just 0 to the failsafe > vlan then the failsafe is broken for this router. > > We could detect that the LAN port is 4 here, but this code is still > not needed for this router. Furthermore, this creates an useless eth0.1 > interface that is visible in eg. ifconfig even when failsafe is not > triggered. This may create problems with bridging since LAN VLAN 1 > would not be passed to the br-lan bridge (and thus to the WLAN, in > the default config) anymore. This will break WLAN VLAN trunking for > instance. if we use board.json then we can simply create 1 lan with eth0 and disable all ports apart from 4. > > Cheers, > Vittorio G > _______________________________________________ > openwrt-devel mailing list > [email protected] > https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel _______________________________________________ openwrt-devel mailing list [email protected] https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
