On 01/09/15 13:00, Simon Wunderlich wrote:
> On Tuesday 01 September 2015 11:00:29 Antonio Quartulli wrote:
>> On 26/08/15 16:41, Simon Wunderlich wrote:
>>> In the case when a temporary entry is added first and a proper tt entry
>>> is added after that, the temporary tt entry is kept in the orig list.
>>> However the temporary flag is removed at this point, and therefore the
>>> purge function can not find this temporary entry anymore.
>>
>> When can this really happen? When a temporary client has roamed to a new
>> originator before it could be claimed by the first one ? Or are there
>> other cases ? I think it is important to make this clear, because the
>> original logic was such as the expected behaviour was to receive an ADD
>> event from the same originator where the temporary client was seen.
>>
>> Other than this, the patch looks good.
> 
> I've seen the problem in that problematic case which was fixed in PATCHv1. 
> Practically, it is unlikely to happen unless there is a malicious attacker or 
> another bug like the one described earlier - that is, speedy join is triggered 
> without any actual TT update.
> 
> The main problem I've perceived here was that the TT entry got the temporary 
> flag removed even if the original sender didn't supply a proper TT 
> announcement yet. The reason for this was that it got a proper reply from 
> another orig node.
> 
> The roaming case you've described would also cause the temporary client to be 
> removed, but might be added later through a ''proper'' TT announcement.
> 
> Do you think I should include any of this in the commit message?

I think you should mention something that makes this case "real": I'd
just say that it is possible that a client detected behind a given
originator moves before an actual claim is made, so triggering this
particular configuration.

I hope you think this is fine.

Cheers,

-- 
Antonio Quartulli