[ovs-dev] [PATCH 0/4] Native tunnel: Update neigh entries in tnl termination.

Paolo Valerio pvalerio at redhat.com
Mon Nov 8 19:28:48 UTC 2021


Gaëtan Rivet <grive at u256.net> writes:

> On Tue, Nov 2, 2021, at 18:12, Paolo Valerio wrote:
>> - Dropped the RFC tag
>> - Included Flavio's suggestions
>> - Squashed patch #5 into #2 and #3
>>
>> The series is composed of the following patches:
>>
>> #1: Expires is modified in different contexts (revalidator, pmd-rx, bfd-tx).
>>     It's probably not very racy for many reasons, but it seems a good idea
>>     to use an atomic here (especially with the introduction of #3).
>>
>> #2: Introduces the possibility to set and read the entries timeout.
>>     It's useful for debugging purpose. Includes a self-test.
>>
>> #3: Fixes a use case in which the tnl neigh entry never
>>     gets updated (further details in the commit message).
>>     Includes a self-test.
>>
>> #4: While working on #2 and #3 I noticed unstable timeouts (more 
>>     visible the ageing time is low) during nd snoop. 
>>     It turned out to be a problem of revalidation. Additional
>>     details can be found in the commit description.
>>
>> The use case fixed in #3 could be fixed relaxing the check in
>> tnl_arp_snoop(), allowing to snoop ARP requests as well, but such
>> approach could add unneeded entries in the cache.
>>
>
> A more general comment on naming.
> I see 'ageing' used seldomly in rstp-related code.
> Otherwise in MAC learning it's 'aging'.
>
> 'ageing' is british English, 'aging' would be correct both in american and british English.
> I am not certain that 'aging' should be preferred but I'm guessing it is.
>

Thank you Gaetan for the feedback.

I picked 'ageing' mostly because it is used in the linux bridge. I
suppose that both are ok.
Nevertheless, Considering that we already use it for MAC learning,
'aging' is probably a better choice.



More information about the dev mailing list