mbox series

[maint,v5,0/3] batman-adv: mcast: BLA fixes

Message ID 20200914195347.10505-1-linus.luessing@c0d3.blue
Headers show
Series batman-adv: mcast: BLA fixes | expand

Message

Linus Lüssing Sept. 14, 2020, 7:53 p.m. UTC
These are three fixes for issues which occur when using the batman-adv
multicast-to-unicast feature.

They fix issues when using the multicast-to-unicast conversion
while BLA is enabled and some nodes are sharing the same LAN side. Here
it either causes "just" duplicates in the "good" scenario (Patch 3/3).
But can also cause multiple BLA backbones to send a frame from the mesh
into the same, shared LAN segment (Patch 2/3). Or in the worst case, even
reflect packets back to the host in the shared LAN, which completely
confuses switches/bridges and ICMPv6 Neighbor Discovery (Patch 1/3).

Regards, Linus

---

Changelog v5:
* Patch 3/3:
  - added missing "*" in kernel doc for batadv_bla_check_duplist()

Changelog v4:
* Patch 1/3:
  - fixed build with CONFIG_BATMAN_ADV_MCAST=n
    -> dummy for batadv_mcast_forw_send_orig()
* Patch 2/3:
  - fixed parameters for batadv_bla_rx() with CONFIG_BATMAN_ADV_BLA=n
    (bool is_bcast -> int packet_type)

Changelog v3:
* removed 1/4, as it was already applied
* renamed title (excluded TT, as already applied)

* Patch 1/3:
  - moved bla-backbone check into multicast code:
    -> to stay on the safe side for net, to touch less other code
       paths - the issue was observed specifically for multicast
       and the fix was tested for that
  - added a note regarding implications for mcast-fanout
* Patch 2/3:
  - previous approach was broken, it would break the DHCPv6 gateway
    code; instead distinguish by batadv 3 vs. 4 addr unicast header
  - added some more code comments
  - updated commit message with new approach
* Patch 3/3:
  - made code a bit more verbose and explicit
  - added some more code comments
  - added missing kernel doc for "orig" parameter to
    new batadv_bla_check_duplist()

Changelog v2:
* Adding "Fixes:" lines