Re: IPv6 NS looback avoidance - RFC7527 support

[Ryota Ozaki <ozaki-r%netbsd.org@localhost>]

On Wed, Mar 7, 2018 at 6:20 PM, Roy Marples <roy%marples.name@localhost> wrote:
> On 07/03/2018 02:54, Ryota Ozaki wrote:
>>
>> On Tue, Mar 6, 2018 at 7:58 PM, Roy Marples <roy%marples.name@localhost> wrote:
>>>
>>> On 02/03/2018 12:37, Roy Marples wrote:
>>>>
>>>>
>>>> Following on from my prior email, this new patch removes the limitation
>>>> of
>>>> only checking the most recent nonce sent.
>>>> It can store 3 nonces (configurable at compile time) and if the user
>>>> wants
>>>> to send more probes, it will store the last 3.
>>>> Currently we default to sending 1 probe and the storage for an extra 2
>>>> is
>>>> just a nice to have.
>>>
>>>
>>>
>>> This has now been committed to -current.
>>> Hopefully someone can test it and we can pull it up to -8.
>>
>>
>> Please write some ATF tests for the change :)
>
>
> I asked how and got no response :)

Oh sorry, I missed that part.

> My best guess would be to write a small program to loopback the NS with and
> without the nonce. Can we do that in ATF? is there an example?

Well, there is no simple way AFAIK.

But this reminds me that I was shown a patch, by k-goda@IIJ, that inserts
packets extracted from a pcap file into a bus of rump shmif. The feature
would help to treat arbitrary packets like this case.

BTW how did you test the feature?

  ozaki-r