to_str.{c,h}. Resolved strange situation where ipx_addr_to_str was
declared in packet.h but defined in packet-ipx.c by moving
ipx_addr_to_str, ipxnet_to_str_punct, and ipxnet_to_str from packet-ipx.{c,h} to to_str.{c,h}
svn path=/trunk/; revision=3219
top-level protocol tree items, use "offset" to set the length, rather
than fetching the length of the item - "offset" is the offset from the
beginning of the Frame Relay or PPP header just past the last field
processed in that header, so it indicates how much of the header we've
processed.
svn path=/trunk/; revision=3215
Encapsulation", not "IETF Encapsulation" - RFC's 1490 and 2427 include
stuff from that appendix, but say that Annex E is the authoritative
source for that information.
Include the second byte of the Cisco HDLC type in the top-level protocol
tree item.
svn path=/trunk/; revision=3213
respectively, not Q.931 and Q.2931, in Frame Relay.
When dissecting Q.933-style multiprotocol encapsulated Frame Relay
frames, use the "osinl" dissector table to check for OSI network layer
protocols, include the NLPID in the tvbuff you hand to
"dissector_try_port()" with that dissector table, and put the NLPID into
the protocol tree as an invisible item - the NLPID is considered part of
the PDU for those protocols, so you have to include it in the tvbuff,
and the dissector will put it into the protocol tree.
Also, make sure the top-level entry for the Frame Relay protocol
includes all the bytes preceding the payload, and none of the payload
bytes.
Export a routine to do Q.933-style dissection, and have the WCP
dissector call it, rather than duplicating that code in the WCP
dissector.
Don't register OSI network layer protocols with the "fr.ietf" dissector
table; it's now sufficient to register them with the "osinl" dissector
table, as the Frame Relay dissector now checks that.
Get rid of unnecessary checks for protocols being enabled (if the
dissector is always called through handles or dissector tables, the
common code for handles and dissector tables will do the checks for
you).
Get rid of some unnecessary #includes.
svn path=/trunk/; revision=3211
"ppp_hdlc" dissector, and call the dissector for "raw" PPP (just RFC
1661, no HDLC encapsulation) the "ppp" dissector.
Have the common routine used by both those dissectors take the offset in
the tvbuff of the PPP protocol field as an argument, rather than
assuming that the protocol field begins at the beginning of the tvbuff,
so we don't have to construct a new tvbuff in the
PPP-in-HDLC-like-framing dissector.
Use the PPP dissector, not the PPP-in-HDLC-like-framing dissector, for
PPP over Frame Relay - there's no HDLC header in PPP over Frame Relay,
at least according to
http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/120newft/120t/120t1/pppframe.htm
svn path=/trunk/; revision=3208
protocols; this obviates the need for "dissect_ppp()" to look at the
protocol field at all.
Have "dissect_ppp_stuff()" add the length of the protocol field to the
length of the top-level protocol tree item, as only it knows how long
the field is.
Have "dissect_mp()" call "dissect_ppp_payload()", rather than
duplicating what "dissect_ppp_payload()" does itself.
Don't fetch the address and control fields in "dissect_ppp()" until
you're ready to put them into the protocol tree.
svn path=/trunk/; revision=3207
"proto_tree_add_item_hidden()", to add the "checksum bad" flags to
packets; the value should be "TRUE", not the numerical value of the
checksum field.
svn path=/trunk/; revision=3202
containing OSI transport layer PDUs).
Enable the Q.931-inside-TPKT code (but not the H.225 stuff, as that
requires Andreas Sikkema's H.225 dissector). Update it to match his
current modified Q.931 dissector.
svn path=/trunk/; revision=3199
"gtk/main.c" in Ethereal.
Add the GLib version to it in Ethereal, and put in the GLib version
rather than the GTK+ version in Tethereal (which isn't linked with
GTK+...).
Make it a GString; this makes the code to construct it slightly less
ugly, especially now that we're putting the GLib version in.
Fix the code for the "-D" flag in Tethereal to compile in a no-libpcap
version (in a no-libpcap version, it just says that this version of
Tethereal wasn't compiled with capture support).
svn path=/trunk/; revision=3196
prints a list of all network interfaces it found on which it can capture
(the same list as the one that shows up in the "Interface" combo box in
Ethereal's "Capture Preferences" dialog).
svn path=/trunk/; revision=3194
"gtk/main.c", and declared in "gtk/gtkglobals.h".
The last argument to "gtk_object_set_data()" is a "gpointer"; cast it to
that.
svn path=/trunk/; revision=3186
has an API that depends on GTK+. "set_main_window_name()" is
UI-toolkit-independent. Declare the former in a new "gtk/ui_util.h"
file, rather than in "ui_util.h"; this helps separate
UI-toolkit-independent stuff from UI-toolkit-dependent stuff.
svn path=/trunk/; revision=3181
as, on a large capture, it could take a significant amount of time.
Let the user stop the computation and, if they do, don't pop up the
statistics dialog box.
Create a new header file declaring the routines to create, update, and
destroy progress dialog boxes; those routines' APIs don't depend on
GTK+, but others declared in "ui_util.h" do, and we don't want to oblige
a source file to depend on GTK+ headers unless it uses a GTK+ API or an
API that depends on GTK+.
svn path=/trunk/; revision=3179
this, as
1) we still need to handle platforms that don't support 64-bit
integral data types, so we still needed the old stuff in some
fashion anyway
and
2) MSVC appears to treat structures as requiring 8-byte
alignment in some cases, and "guint64"s require 8-byte
alignment on at least some platforms, forcing structures
containing those 64-bit time stamps to have a size that's a
multiple of 8 bytes, which *isn't* the correct size for the
data record header.
svn path=/trunk/; revision=3177
Ed Warnicke