The code filling the fp_info structure does not know the UDP Port / IP
Address pairs, no. I realize that this would make things easier.
In the case of FP being sent via Ethernet, the FP frames are carried
within UDP. Port assignment is arbitrary, that is, the first UDP layer
encountered always contains FP, the 'upper' UDP layer contains the
payload. At least this is my understanding how things work.
Concerning the attached trace: The problem is that for correctly
dissecting the FP frames, additional information not necessarily
available from the packets contained in the trace are required (e.g.
detecting which type of FP channel is contained within a certain
connection).
Thanks!
Tobias
Anders Broman wrote:
> Hi,
> I'm not intierly sure how this is supposed to work but my understanding
> is that one of the heuristic
> Conditions is that the "fp structure" is filled in/attatched to the
> packet.
>
> Does the code filling in this structure know the IP address and port
> pair going to be used for the UDP transport?
> If so that code could set up a conversation and set the UDP conversation
> dissector.
> That means that for that UDP IP and port par "dissect_fp..() would be
> called directly not the heuristic.
> I asume that any IP payload traffic carried inside of this UDP reaffic
> would have other IP address port combinations?
> With this setup there is no need for a FP heuristic dissector at all.
>
> As for thye other question regarding the trace attached to the bug is
> there enough iformation in the (NBAP?)
> Message(s) to fill in the fp struct if so code to do that could be added
> to the NBAP dissector + setting up of a conversation.
>
> Regards
> Anders
>
>
> -----Original Message-----
> From: wireshark-dev-bounces@xxxxxxxxxxxxx
> [mailto:wireshark-dev-bounces@xxxxxxxxxxxxx] On Behalf Of Tobias Witek
> Sent: den 11 november 2009 11:33
> To: wireshark-dev@xxxxxxxxxxxxx
> Subject: [Wireshark-dev] Dissecting UDP conversations that encapsulate
> UDPdata
>
> Hello,
>
> I am currently trying to extend the UMTS FP protocol to also handle FP
> frames sent via UDP. To avoid modifications in the UDP dissector, my
> plan was to use a heuristic to detect the first UDP packet of each
> stream that contains FP and assign a conversation dissector to the whole
> conversation.
>
> The problem is that these frames can in turn contain UDP (e.g. DNS
> queries). The solution outlined above would dissect these as FP again,
> if I understand correctly (as a separate conversation).
>
> As far as I see, I would need a way to let the heuristic determine that
> the current frame was already parsed as FP (respectively, is already
> part of a different UDP conversation) and should not be treated as FP.
>
> Currently, I see no way to do this and would be very happy about any
> suggestions!
>
> Regards,
>
> Tobias
> ________________________________________________________________________
> ___
> Sent via: Wireshark-dev mailing list <wireshark-dev@xxxxxxxxxxxxx>
> Archives: http://www.wireshark.org/lists/wireshark-dev
> Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev
>
> mailto:wireshark-dev-request@xxxxxxxxxxxxx?subject=unsubscribe
> ___________________________________________________________________________
> Sent via: Wireshark-dev mailing list <wireshark-dev@xxxxxxxxxxxxx>
> Archives: http://www.wireshark.org/lists/wireshark-dev
> Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev
> mailto:wireshark-dev-request@xxxxxxxxxxxxx?subject=unsubscribe