Hi Harald,
On Sun, Dec 06, 2009 at 12:29:40PM +0530, Harald Welte wrote:
Dear Jai,
thanks for your work.
On Fri, Dec 04, 2009 at 05:10:00PM +0530, Jai Menon wrote:
Attached is a direct port of the IML dissector
code from the ip.access
code to wireshark SVN head with a few changes to make it work with the
current API. This was done as part of Harald's GSM workout during
FOSS.IN and he suggested I post it to this list. I haven't really
tested this code since I dont have access to IML traces, but assuming
the original code worked, this one should too.
Great. Anyone who has two nanoBTS can set up a multi-TRX BTS and
trace the IML protocol between those two BTS. This pcap should be possible
to decode using this dissector.
Also, if someone could upload/post the resulting pcap, that would be
useful for local testing.
Before we can submit this mainline, we need to clean
up packet-gsm.h though.
ip.access has one header file (packet-gsm.h) where they define many aspects of
various different GSM protocols. This is not really how things are done
in upstream wireshark.
Thus, we need to remove everything that is not IML related, like
[...]
etc. I could now continue for the whole file, but I
think you
get what I mean :)
Indeed :) the first attempt was quite naive as in I wanted to just get it
to build cleanly with the new API.
Would you be able to clean this further up and do
another round of patch
submission? Thanks again for your help.
Sure thing. Before I go ahead and do that, I wanted to know whats the
preferred approach here. The packet-gsm_ipa dissector code, for
example, doesn't #include any additional headers. Should the iml code
be modified so that all typedefs like ie_def_t are present in the one
source file? Or should the packet-gsm.h header be trimmed down to
contain only what the iml code requires? If it is cleaned up to remove
unused stuff, should the name then be changed to packet-gsm_iml.h or
something similar?
Thanks for the comments.
--
Jai Menon