Lars Dörner wrote:
It's a RFB based protocol running on TCP.
To dissect the PDU I need the length of it, put the Problem is The length is
encoded at the end of the PDU.
Presumably there is *something* in the data stream that indicates where
the PDU ends - i.e., it *has* to be possible to scan forward in the data
stream to find the end of the PDU.
Or is there a possibility to continue with the Next TCP packet if I Realize
the end of the Video PDU is not reached?
Yes. You'd need to do TCP reassembly, but you'd process the data in the
current TCP segment - *without* building the protocol tree - and:
if the PDU ends within the TCP segment, dissect it, and then start
processing the next PDU (I'm assuming the next PDU is immediately after
the previous one);
if the PDU *doesn't* end within the TCP segment, set
pinfo->desegment_offset to the offset within the tvbuff of the first
byte of that PDU, and set pinfo->desegment_len to 1 (which requests that
at least one more byte be added to the PDU from the next segment), and
then return.
See, for example, the code that dissects the entity headers in
"req_resp_hdrs_do_reassembly()".
