https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=16470
--- Comment #14 from Jaap Keuter <jaap.keu...@xs4all.nl> ---
(In reply to levente.meszaros from comment #11)
> I agree with your analysis. With my last comment I only wanted to note that
> if the interface is not looked at when dissecting the conversation, then the
> dissector may be confused again by completely unrelated frames on other
> interfaces.
Yes, but this is a more generic problem which comes up at every layer in the
protocol stack. Image different VLANs, Provider bridges, MPLS labels, IP-in-IP,
tunnels of all kinds, etc.
Currently there are limited features incorporated to handle this, the so called
conversation mechanism. It basically does this by means of endpoint
identification, which started with TCP and UDP connections, but is being
stretched to cover other endpoint types as well. Unfortunately this design can
only bring us so far, to create a solid solution a partial redesign of the
dissection engine is needed, to create the concept of context, starting at the
lowest layer (phy).
BTW: current solution works by the grace of the interface direction (In/Out)
being present for the frames in the capture file. If not then this won't work
either.
--
You are receiving this mail because:
You are watching all bug changes.
___________________________________________________________________________
Sent via: Wireshark-bugs mailing list <wireshark-bugs@wireshark.org>
Archives: https://www.wireshark.org/lists/wireshark-bugs
Unsubscribe: https://www.wireshark.org/mailman/options/wireshark-bugs
mailto:wireshark-bugs-requ...@wireshark.org?subject=unsubscribe
