Linux Ethernet Bridging - May 2021 - [Bridge] [PATCH net-next v3 00/11] net: bridge: split IPv4/v6 mc router state and export for batman-adv

On 13/05/2021 02:19, Linus L?ssing wrote:
> Hi,
> 
> The following patches are splitting the so far combined multicast router
> state in the Linux bridge into two ones, one for IPv4 and one for IPv6,
> for a more fine-grained detection of multicast routers. This avoids
> sending IPv4 multicast packets to an IPv6-only multicast router and 
> avoids sending IPv6 multicast packets to an IPv4-only multicast router.
> This also allows batman-adv to make use of the now split information in
> the final patch.
> 
> The first eight patches prepare the bridge code to avoid duplicate
> code or IPv6-#ifdef clutter for the multicast router state split. And 
> contain no functional changes yet.
> 
> The ninth patch then implements the IPv4+IPv6 multicast router state
> split.
> 
> Patch number ten adds IPv4+IPv6 specific timers to the mdb netlink
> router port dump, so that the timers validity can be checked individually
> from userspace.
> 
> The final, eleventh patch exports this now per protocol family multicast
> router state so that batman-adv can then later make full use of the 
> Multicast Router Discovery (MRD) support in the Linux bridge. The 
> batman-adv protocol format currently expects separate multicast router
> states for IPv4 and IPv6, therefore it depends on the first patch.
> batman-adv will then make use of this newly exported functions like
> this[0].
> 
> Regards, Linus
> 
> [0]:
https://git.open-mesh.org/batman-adv.git/shortlog/refs/heads/linus/multicast-routeable-mrd
>      ->
https://git.open-mesh.org/batman-adv.git/commit/d4bed3a92427445708baeb1f2d1841c5fb816fd4
> 
Nice work overall, thank you. I hope it was tested well. :)
It'd be great if later you could add some selftests.

Cheers,
 Nik

On Thu, May 13, 2021 at 03:02:13PM +0300, Nikolay Aleksandrov
wrote:
> Nice work overall, thank you. I hope it was tested well. :)
> It'd be great if later you could add some selftests.
> 
> Cheers,
>  Nik
Hi Nikolay,

I think I found a way now to better deal with the protocol
specific hlist_for_each_entry(), by using hlist_for_each()
and a helper function, to reduce the duplicate code
with br_{ip4,ip6}_multicast_get_rport_slot() as you suggested
(and also removing duplicate code with 
br_{ip4,ip6}_multicast_mark_router()) and reworked patches 7
and 9 a bit for that...

Sorry for the inconvience and my bad timing with your reviews. But
thanks a lot for all your valuable feedback!

Also netdevbpf patchwork had a few more remarks, they should
hopefully be fixed now, too.

Regards, Linus