[Nfd-dev] PIT matching exclude question

[Junxiao Shi]

Hi Sonu

Can you provide tcpdump (not ndndump) trace from the node in question?
Capture one trace on every interface used in the experiment.
If some traffic is from an application on that node, force the application
to use TCP transport (set NDN_CLIENT_TRANSPORT=tcp4://127.0.0.1:6363
environ), and tcpdump at lo interface.

Also, all bug reports should be sent to nfd-dev mailing list, not
individual developers.
Have a look at http://www.lists.cs.ucla.edu/pipermail/nfd-dev/2016-May/
001748.html on how to write a bug report effectively.

Yours, Junxiao

On Wed, Dec 21, 2016 at 9:05 AM, Lixia Zhang <lixia at cs.ucla.edu> wrote:

>
>
>
> Suppose a node *x*, which has received data A in round 3, has a data
> interest *m*: data/interest/prefix/03.exclude<A> outstanding. It then
> receives a new data interest *n:* data/interest/prefix/03.exclude<A,B>
> from another node *y*. The RoundSync at node *x* sends data A with name
> data/interest/prefix/03 to satisfy the interest *n.* However the NFD also
> marks the outstanding interest *m* as satisfied because it has the same
> name. Now if a legitimate data B arrives at node *x*, which could satisfy
> interest *m*, the NFD discards the data.
>
>
> I copied Junxiao and Beichuan on this reply: the above seems an
> implementation bug: the PIT entry management considered the name matching
> only, but in this case one must consider the exclude filter, otherwise it
> does the wrong thing (e.g.your question-1)
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/nfd-dev/attachments/20161221/aaf19e38/attachment-0001.html>