(removed in r48218) which did nothing particularly useful. Also lets us remove
another debugging environment variable.
svn path=/trunk/; revision=48219
glib memory slices.
- We weren't doing anything with the emem slab that couldn't be done with glib
slices.
- Removes a fair bit of code as well as one debugging environment variable.
- Glib slices are much cache-friendlier and are multi-threading friendly (if
we ever go there).
- Allows glib to actually return slices to the OS on occasion. The emem slab
would hold onto its memory forever which resulted in a great deal of wasted
memory after closing a large file.
svn path=/trunk/; revision=48218
Don't add an extra 16 to offset, we've been updating it as we go. Fixes cases
where valid AMS packets were marked as malformed.
svn path=/trunk/; revision=48215
Added functionality:
- SMB2 support for Export->Objects->SMB
- support for SMB_COM_CREATE, SMB_COM_OPEN, SMB_COM_READ and SMB_COM_WRITE commands
- Ability to choose between File Id and full file name as identifier for file re-building. Implemented as an option under Edit->Preferences->Protocols->SMB and Edit->Preferences->Protocols->SMB2.
Other minor changes and fixes:
- Full filename in file
- Inclusion of IP of SMB server when treeid name (i.e. hostname) is not known
- UTF-8 filenames encoding before passing them to Export Object Window
- Re-written insert_chunk function of export_object_smb.c to make it easier to debug
- Fixed of an error in insert_chunk function of export_object_smb.c (the verification of next free_chunk was always skipped after deleting one free_chunk).
- Removed duplicated code by inserting the function feed_eo_smb in packet-smb.c and packet-smb2.c
- Changed the label of Export->Objects->SMB menu into Export->Objects->SMB/SMB2
svn path=/trunk/; revision=48210
1. Convert more proto_tree_add_text to proto_tree_add_<something else>/expert_info. checkAPIs.pl is happy (for packet-rtps.c), but the raw number is a bit deceiving because of macros and wrappers.
2. Fixed the offending "uses snprintf + strlen to assemble strings" in packet-rtps.c. The exact same code is used in packet-rtps2.c, so just proper refactoring will fix it. There is still too much unnecessary use of g_snprintf/g_strlcpy/strlen, but that's for a later date.
3. Removed most of the "useless" wrapper functions. Again, the number of proto_tree_add_text is deceiving, so the number of hfs that really need to be created is much larger.
4. Whitespace cleanup. Removed a lot of whitespace so I could see more code on the screen, to help determine duplication between packet-rtps.c and packet-rtps2.c Comments/descriptions of fields remain untouched.
The more I trim, the more I think this should all be in a single dissector file, which will be the goal of the next update. Trying to patch this in somewhat manageable chunks.
Also need to submit sample traces generated for (fuzz)testing.
svn path=/trunk/; revision=48206
a theme-specific icon (and as such hicolor is where applications should
install their icons). So: don't install some of our icons in the gnome area,
install them all in hicolor.
While we're at it, go ahead and install all the icon sizes we have.
If we're on SuSE, use their desktop-file-updater macro; without that they
won't recognize our desktop file.
Fix bug which prevented the MIME database from being updated if our install
prefix is not /usr .
svn path=/trunk/; revision=48204
adding the fields to the protocol tree as we fetch them, rather than
fetching a bunch of them up front. That way, if the packet was cut
short by a snapshot length, we'll dissect what we have.
Create the top-level tree item at the beginning, with an unknown length,
and set the length when we're done. If we're putting details into that
item, update the item as we process each field; also update the Info
column as we process each field.
Don't use tvb_bytes_exist() to do our own tvbuff bounds-checking; let
the tvbuffs themselves do that. Do not use it to do header-length
checking; check the actual header length value instead.
Do not fail if the *captured* data in the tvbuff is too short; we're
selected by an IP protocol number, which we can expect to definitively
identify us, not by a transport-layer port number, which is often an
unreliable identifier.
Do header length checks as we go along, and bail as soon as we identify
the header length as wrong. (We do the first check once we get the X
bit, so we know whether the generic header is 12 or 16 bytes long.)
Treat a too-*large* header as a protocol violation, not a malformed
packet indication.
Use tvb_get_ntoh24() and tvb_get_ntoh48() to fetch the sequence number,
rather than fetching it in pieces and putting them together ourselves.
Correctly pluralize "byte".
Don't use tvb_length_remaining() to check whether we have a payload, use
tvb_reported_length_remaining(), so we base it on whether the packet
actually had the data, not on whether we actually captured it.
svn path=/trunk/; revision=48201
If we don't decode a tag, just say "Undecoded" on the top-level item;
the tag name or number is already on that item, and the length is
underneath it.
svn path=/trunk/; revision=48180
[PATCH 1/2] Revert "Try to fix the "LNK4217: locally defined symbol"
warnings.
This reverts commit r48158.
[PATCH 2/2] Employ small hack in editcap to link with a few objects from
libwireshark properly
From me:
Add the ability to reset symbol exports via ws_symbol_export.h's include
guard and do so in capinfos.c and editcap.c. We include ws_symbol_export.h
in over 200 files so it didn't seem to make sense to remove its include
guard entirely.
svn path=/trunk/; revision=48170
Michael Tüxen