If PYTHON_DIR is specificed in config.nmake the variable will be overritten by
PYTHON_DIR=C:\Python$(PYTHON_VER).
It looks like the problem was created by revision 34246.
svn path=/trunk/; revision=34673
The rest of the code assumes that sip can be NULL, so don't assert when it's
not.
Also make fid_cmp() static since it's only used in this module.
svn path=/trunk/; revision=34663
As mentioned before, the heuristics used by packet-adwin-config were weak for
UDP packets. This patch creates a stronger heuristic by checking if one of the
MAC addresses in the UDP packets is in the range assigned to the manufacturer
of the devices that use this protocol (or if it is a broadcast).
That should be relatively safe to do because the protocol is rarely used in
routed networks.
svn path=/trunk/; revision=34654
You may not use g_int64_hash() or g_int64_equal(), as they are not present in the minimum version of GLib that we support. Create a ieee802154_long_addr structure for long addresses, create hash routines for those addresses, and use them.
Export and use those routines.
svn path=/trunk/; revision=34653
present in the minimum version of GLib that we support. Create a
ieee802154_long_addr structure for long addresses, create hash
routines for those addresses, and use them.
svn path=/trunk/; revision=34651
Attached patch:
1. Adds port 5985 as a HTTP traffic port (used by MS Powershell remoting over
HTTP)
2. Adds dissection of Kerberos authentication to HTTP.
svn path=/trunk/; revision=34641
The company I work for uses two proprietary protocols, for which I initially
developed wireshark plugins. Now we would like to integrate them into the
public wireshark repository.
I followed the READMEs and converted the plugins into a static dissectors. I
cleaned up the code until checkAPI.pl was silent, translated all terms to
english and ran randpkt and fuzz-testing for a long time. All that I found was
a bug in a different dissector.
From me:
- Fold the header files into the dissectors
- Clean up some memory leaks
- Strengthen the heuristics of adwin-config (the TCP heuristics are still pretty
weak)
- Make packet-adwin.c a "new style" dissector
- Use find_or_create_conversation()
- Remove most of the check_col()'s
svn path=/trunk/; revision=34640
BACnet has a private transfer service which is vendor specific. The start of
each request and response contains the vendor identifier. I've added a way for
vendors to provide their own dissectors by registering their vendor identifier.
The packet-bacapp.c method fConfirmedPrivateTransfer has been modified to look
for a vendor specified dissector. If found it will be run. If not found we
default to running the standard dissection included in packet-bacapp.c.
I modified the summary column display for private transfer messages so that the
summary now displays the Vendor Identifier (V=xx) and the Service Number (SN=xx).
https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=5250
From me: Rename sub-dissector tablle to "bacapp.vendor_identifier"
Change subdissector ui_name to "BACapp Vendor Identifier"
svn path=/trunk/; revision=34625
Find a patch who clean up PPP dissector
* Remove check_col (from my previous patch #5325)
* Remove the #define ppp_min (unused)
* Remove some unused ett_*
* Cleanup NO ARRAY entry (it is a false positive of checkhf tool, it is need {
(in the same line) before a hf_... in hf_register_info struct
* Cleanup Unused entry in VSNP Dissector ( it's really strange, there is no
check when the code was added #4434 ;) )
* minor bug fix...
From me: put one of the check_col()'s (that also protects a val_to_str() call)
back in.
Note: the checkhf tool bug was fixed in rev 34623.
svn path=/trunk/; revision=34624
is on the preceding line. There was code here to do it but it needed to be
moved inside the preceding 'else' (in particular, before the 'next' sent us
around the loop again).
The bug was noticed by Alexis La Goutte in https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=5329
svn path=/trunk/; revision=34623