Wireshark · Ethereal-dev: Re: [Ethereal-dev] Cleanup function for each frame

[Stig Bjørlykke <stig@xxxxxxxxxxxxx>]

On 28. nov. 2005, at 22.13, Ulf Lamping wrote:

> Stig Bjørlykke wrote:
> > I need something like this in COTP in the situation where the  
> > first  COTP segment in a frame is the last of a fragmented block  
> > (the rest  received in previous frames), and the second COTP  
> > segment is a non- fragmented block.  The COTP length is 2, which  
> > means dst_ref is  always 0 (used as "id" in the fragment  
> > functions), and the last COTP  segment is messing up with the  
> > fragment reassembly (because the frame  is dissected twice).
> Is the frame dissected twice, or is the dissector called twice  
> because of the two appearances in this frame?
It looks like the frame is dissected twice, because the dissector is  
called four times when I have two COTP segments in the frame.  I'm  
using the latest svn on Mac OS X.
> > If I can reset a counter after dissecting this  frame this counter  
> > can be used as "id".
> Wouldn't this mess up the reassembling of the following frames?
When I use the attached patch (where I have patched packet-frame.c to  
do what I want) it appears to work correctly, also for following  
frames, but I don't know if this is "correct coding"?
> BTW: do you have an example capture?
I will make a public releasable capture available tomorrow, but I can  
see the same thing happening in p772-transfer-success.pcap from  
http://wiki.ethereal.com/S4406 - look at frame 20.

--
Stig Bjørlykke

Attachment: packet-clnp.patch
Description: Binary data