Title: RE: New dissector for SNDCP protocol
Ouch. I shouldn't have done development on two different machines. Attached is the latest version of the SNDCP dissector, please disregard the attachment in my previous message below.
Christian
-----Original Message-----
From: Christian Falckenberg [mailto:christian.falckenberg@xxxxxxxxxxxxxxxxxx]
Sent: Montag, 12. April 2004 19:38
To: ethereal-dev@xxxxxxxxxxxx
Subject: New dissector for SNDCP protocol
Hi folks,
after my name has been added to the Ethereal authors list for this
little piece of initial Megaco code, I thought I would have to deliver
some more to justify that :-).
Attached is a new dissector for the SNDCP protocol defined in 3GPP TS
04.65. It is the next protocol on top of the GPRS Logical Link Control
layer posted by Josef Korelus a few days ago, and it is the missing
piece in the protocol tree until we can continue to use existing
dissectors for the IP protocol carried by SNDCP. My main purpose was to
be able to use the existing TCP analysis features for traces captured on
the Gb interface of GPRS. Most of it works already very well, only the
TCP graphs are not detecting TCP on top of this protocol (any idea how
to fix this? BTW, the same problem applies to tcpdump over PPP on Linux;
the TCP graphs don't seem to work for "Linux Cooked Capture")).
The dissector has been tested only with an existing GPRS Gb trace for
unacknowledged data without any compression. The protocol also supports
acknowledged data transfer, and the code for it has been added, but I
would need an example trace to verify that it is working in this mode as
well. If data and/or protocol compression is used, then I simply hand
off to the data dissector instead of IP. Each frame only contains a
pointer to a compression mechanism negotiated before, so that it will be
very difficult if not impossible to decompress the data.
I do already support the reassembly of N-PDUs, thanks to the procedures
in reassemble.c and to the good existing example code.
The original input trace with the SNDCP protocol was captured by a
Tektronix K1205. Unfortunately, there is no support for this analyzer
yet (anyone planning to write this? If not, I might try it), and I used
a small perl script to convert the ASCII format of that trace to
libpcap. This script is based on the rf52pcap script posted by Josef
Korelus a while ago, but I would rather see real support for the K12
instead of using and distributing this script.
Other than with the Megaco code which was only a proof-of-concept for
me, I wouldn't mind continuing with the development of this SNDCP
dissector. Any suggestions for improvements, bug reports etc. are welcome.
Christian
Attachment:
packet-sndcp.c
Description: Binary data