Wireshark · Ethereal-dev: Re: [Ethereal-dev] Need help with protocol that spans multiple TVBs

[Guy Harris <gharris@xxxxxxxxx>]

On Sun, Apr 11, 2004 at 03:47:00PM -0400, Herbert Falk wrote:
> I have attached an image that shows my particular problem.  The square 
> shows that the MMS PDU spans three(3) TVBs.  I have re-assembly turned 
> on at TCP, TPKT, and COTP.  When the MMS dissector is called, the 
> function is only given the MMS information in the first packet.
> 
> MMS uses Session and Presentation on-top of COTP.

Presumably you're using the Session and Presentation dissectors that
come with Ethereal?

At what layer is the fragmentation in question being done? The TPKT
dissector supports reassembly; the COTP dissector also supports it,
*but* it's buggy (the reassembly code currently doesn't have good
support for transport-layer protocols, i.e.  protocols where, instead of
the fragments of a packet having sequence numbers 0 through N-1 or 1
through N or something such as that, they have sequence numbers M
through M+N-1, i.e.  where the sequence number is a transport-layer
sequence number over the entire connection rather than a fragment
sequence number); the session dissector also supports it - is there any
segmentation/reassembly that would be done at the presentation layer?