Linux Ethernet Bridging - Sep 2022 - [Bridge] [PATCH v5 net-next 6/6] selftests: forwarding: add test of MAC-Auth Bypass to locked port tests

Sorry for the delay, was away.

On Tue, Sep 27, 2022 at 10:33:10AM +0200, netdev at kapio-technology.com
wrote:
> On 2022-09-21 09:15, Ido Schimmel wrote:
> > 	bridge fdb add `mac_get $h2` dev br0 blackhole
> 
> To make this work, I think we need to change the concept, so that blackhole
> FDB entries are added to ports connected to the bridge, thus
>      bridge fdb add MAC dev $swpX master blackhole
> 
> This makes sense as the driver adds them based on the port where the SMAC
is
> seen, even though the effect of the blackhole FDB entry is switch wide.
Asking user space to associate a blackhole entry with a bridge port does
not make sense to me because unlike regular entries, blackhole entries
do not forward packets out of this port. Blackhole routes and nexthops
are not associated with a device either.
> Adding them to the bridge (e.g. f.ex. br0) will not work in the SW bridge
as
> the entries then are not found.
Why not found? This works:

 # bridge fdb add 00:11:22:33:44:55 dev br0 self local
 $ bridge fdb get 00:11:22:33:44:55 br br0
 00:11:22:33:44:55 dev br0 master br0 permanent

With blackhole support I expect:

 # bridge fdb add 00:11:22:33:44:55 dev br0 self local blackhole
 $ bridge fdb get 00:11:22:33:44:55 br br0
 00:11:22:33:44:55 dev br0 master br0 permanent blackhole

On 2022-09-28 08:59, Ido Schimmel wrote:
> Sorry for the delay, was away.
Good to have you back. :-)
> 
> On Tue, Sep 27, 2022 at 10:33:10AM +0200, netdev at kapio-technology.com 
> wrote:
>> On 2022-09-21 09:15, Ido Schimmel wrote:
>> > 	bridge fdb add `mac_get $h2` dev br0 blackhole
>> 
>> To make this work, I think we need to change the concept, so that 
>> blackhole
>> FDB entries are added to ports connected to the bridge, thus
>>      bridge fdb add MAC dev $swpX master blackhole
>> 
>> This makes sense as the driver adds them based on the port where the 
>> SMAC is
>> seen, even though the effect of the blackhole FDB entry is switch 
>> wide.
> 
> Asking user space to associate a blackhole entry with a bridge port 
> does
> not make sense to me because unlike regular entries, blackhole entries
> do not forward packets out of this port. Blackhole routes and nexthops
> are not associated with a device either.
> 
>> Adding them to the bridge (e.g. f.ex. br0) will not work in the SW 
>> bridge as
>> the entries then are not found.
> 
> Why not found? This works:
> 
>  # bridge fdb add 00:11:22:33:44:55 dev br0 self local
>  $ bridge fdb get 00:11:22:33:44:55 br br0
>  00:11:22:33:44:55 dev br0 master br0 permanent
> 
> With blackhole support I expect:
> 
>  # bridge fdb add 00:11:22:33:44:55 dev br0 self local blackhole
>  $ bridge fdb get 00:11:22:33:44:55 br br0
>  00:11:22:33:44:55 dev br0 master br0 permanent blackhole
In my previous replies, I have notified that fdb_find_rcu() does not 
find the entry added with br0, and thus fdb_add_entry() that does the 
replace does not replace but adds a new entry. I have been thinking that 
it is because when added with br0 as dev it is added to dev br0's fdb, 
which is not the same as 'dev <Dev> master' fdb...

I think bridge fdb get works in a different way, as I know the get 
functionality gets all fdb entries from all devices and filters them (if 
I am not mistaken)...

On 2022-09-28 08:59, Ido Schimmel wrote:
> Why not found? This works:
> 
>  # bridge fdb add 00:11:22:33:44:55 dev br0 self local
>  $ bridge fdb get 00:11:22:33:44:55 br br0
With:
  # bridge fdb replace 00.11.22.33.44.55 dev $swpX static

fdb_find_rcu() will not find the entry added with 'dev br0' above, and 
will thus add a new entry afaik.