"epan/..." pathnames, so as to avoid collisions with header files in any
of the directories in which we look (e.g., "proto.h", as some other
package has its own "proto.h" file which it installs in the top-level
include directory).
Don't add "-I" flags to search "epan", as that's no longer necessary
(and we want includes of "epan" headers to fail if the "epan/" is left
out, so that we don't re-introduce includes lacking "epan/").
svn path=/trunk/; revision=4586
structure to the "packet_info" structure; only stuff that's permanently
stored with each frame should be in the "frame_data" structure, and the
"column_info" structure is not guaranteed to hold the column values for
that frame at all times - it was only in the "frame_data" structure so
that it could be passed to dissectors, and, as all dissectors are now
passed a pointer to a "packet_info" structure, it could just as well be
put in the "packet_info" structure.
That saves memory, by shrinking the "frame_data" structure (there's one
of those per frame), and also lets us clean up the code a bit.
svn path=/trunk/; revision=4370
of protocol-id-plus-datum pairs, so that multiple protocols can attach
information to the same conversation.
Dissectors that attach information to a conversation should not assume
that if they find a conversation it has one of its data attached to it;
the conversation might've been created by another dissector.
svn path=/trunk/; revision=3901
"void *" that a dissector can set to point to such a structure; that
means that the stuff in the epan directory doesn't have to know anything
about the protocol-specific private data one dissector passes to
another, and that structure doesn't have to be changed if a dissector
wants to pass some new type of data to another dissector.
svn path=/trunk/; revision=3818
protocols, in addition to adding structures to the list of filterable
fields. Give it an extra argument that specifies a "short name" for the
protocol, for use in such places as
pinfo->current_proto;
the dialog box for constructing filters;
the preferences tab for the protocol;
and so on (although we're not yet using it in all those places).
Make the preference name that appears in the preferences file and the
command line for the DIAMETER protocol "diameter", not "Diameter"; the
convention is that the name in question be all-lower-case.
Make some routines and variables that aren't exported static.
Update a comment in the ICP dissector to make it clear that the
dissector won't see fragments other than the first fragment of a
fragmented datagram.
svn path=/trunk/; revision=2810
- packet-afs.c: dissect_acl() didn't restrict the size of a string read
with sscanf(). An exploit has been released.
- packet-nbns.c: When passed an illegal name, get_nbns_name() would
overrun nbname with an error message. This isn't exploitable AFAIK,
but it could result in a crash.
- packet-ntp.c: dissect_ntp() wasn't checking the length of the
reference clock's host name. This is most likely exploitable.
This fix simply lops off the end of the host name if it's too long.
We should probably add an ellipsis (...) as we have done in other
places in the code.
svn path=/trunk/; revision=2671
can be put, and a pointer to the string for the column, which might or
might not point to that buffer.
Add a routine "col_set_str()", which sets the string for the column to
the string passed to it as an argument; it should only be handed a
static string (a string constant would be ideal). It doesn't do any
copying, so it's faster than "col_add_str()".
Make the routines that append to columns check whether the pointer to
the string for the column points to the buffer for the column and, if
not, copy the string for the column to the buffer for the column so that
you can append to it (so you can use "col_set_str()" and then use
"col_append_str()" or "col_append_fstr()").
Convert a bunch of "col_add_str()" calls that take a string constant as
an argument to "col_set_str()" calls.
Convert some "col_add_fstr()" calls that take a string constant as the
only argument - i.e., the format string doesn't have any "%" slots into
which to put strings for subsequent arguments to "col_set_str()" calls
(those calls are just like "col_add_str()" calls).
Replace an END_OF_FRAME reference in a tvbuffified dissector with a
"tvb_length(tvb)" call.
svn path=/trunk/; revision=2670
the following:
It is now possible to enable/disable a particular protocol decoding
(i.e. the protocol dissector is void or not). When a protocol
is disabled, it is displayed as Data and of course, all linked
sub-protocols are disabled as well.
Disabling a protocol could be interesting:
- in case of buggy dissectors
- in case of wrong heuristics
- for performance reasons
- to decode the data as another protocol (TODO)
Currently (if I am not wrong), all dissectors but NFS can be disabled
(and dissectors that do not register protocols :-)
I do not like the way the RPC sub-dissectors are disabled (in the
sub-dissectors) since this could be done in the RPC dissector itself,
knowing the sub-protocol hfinfo entry (this is why, I've not modified
the NFS one yet).
Two functions are added in proto.c :
gboolean proto_is_protocol_enabled(int n);
void proto_set_decoding(int n, gboolean enabled);
and two MACROs which can be used in dissectors:
OLD_CHECK_DISPLAY_AS_DATA(index, pd, offset, fd, tree)
CHECK_DISPLAY_AS_DATA(index, tvb, pinfo, tree)
See also the XXX in proto_dlg.c and proto.c around the new functions.
svn path=/trunk/; revision=2267
a particular type, rather than taking a varargs list, along the lines of
the "proto_tree_add_XXX_format()" routines.
Replace most calls to "proto_tree_add_item()" and
"proto_tree_add_item_hidden()" with calls to those routines.
Rename "proto_tree_add_item()" and "proto_tree_add_item_hidden()" to
"proto_tree_add_item_old()" and "proto_tree_add_item_hidden_old()", and
add new "proto_tree_add_item()" and "proto_tree_add_item_hidden()"
routines that don't take the item to be added as an argument - instead,
they fetch the argument from the packet whose tvbuff was handed to them,
from the offset handed to them.
svn path=/trunk/; revision=2031
Add exceptions routines.
Convert proto_tree_add_*() routines to require tvbuff_t* argument.
Convert all dissectors to pass NULL argument ("NullTVB" macro == NULL) as
the tvbuff_t* argument to proto_tree_add_*() routines.
dissect_packet() creates a tvbuff_t, wraps the next dissect call in
a TRY block, will print "Short Frame" on the proto_tree if a BoundsError
exception is caught.
The FDDI dissector is converted to use tvbuff's.
svn path=/trunk/; revision=1939
to a type requiring 2-byte or better alignment and was then
dereferenced; doing that requires that the code generated by your
compiler not trap if it makes an unaligned reference, and on most RISC
processors the code generated by the compiler *will* trap on an
unaligned reference by default.
svn path=/trunk/; revision=1480
routines, which are called before a dissection pass is made over all the
packets in a capture - the "init" routine would clear out any state
information that needs to be initialized before such a dissection pass.
Make the NCP, SMB, AFS, and ONC RPC dissectors register their "init"
routines with that mechanism, have the code that reads in a capture file
call the routine that calls all registered "init" routines rather than
calling a wired-in set of "init" routines, and also have the code that
runs a filtering or colorizing pass over all the packets call that
routine, as a filtering or colorizing pass is a dissection pass.
Have the ONC RPC "init" routine zero out the table of RPC calls, so that
it completely erases any state from the previous dissection pass (so
that, for example, if you run a filtering pass, it doesn't mark any
non-duplicate packets as duplicates because it remembers them from the
previous pass).
svn path=/trunk/; revision=1050
dynamically-assigned "ett_" integer values, assigned by
"proto_register_subtree_array()"; this:
obviates the need to update "packet.h" whenever you add a new
subtree type - you only have to add a call to
"proto_register_subtree_array()" to a "register" routine and an
array of pointers to "ett_", if they're not already there, and
add a pointer to the new "ett_" variable to the array, if they
are there;
would allow run-time-loaded dissectors to allocate subtree types
when they're loaded.
svn path=/trunk/; revision=1043
Replace "add_to_conversation()" with:
"conversation_new()", which creates a new conversation, given
source and destination addresses and ports, and returns a
pointer to the structure for the conversation;
"find_conversation()", which tries to find a conversation for
given source and destination addresses and ports, and returns a
pointer to the structure for the conversation if found, and a
null pointer if not found.
Add a private data pointer field to the conversation structure, and have
"conversation_new()" take an argument that specifies what to set that
pointer to; that lets clients of the conversation code hang arbitrary
data off the conversation (e.g., a hash table of protocol requests and
replies, in case the protocol is a request/reply protocol wherein the
reply doesn't say what type of request it's a reply to, and you need
that information to dissect the reply).
svn path=/trunk/; revision=920
structure to "dl_src"/"dl_dst", "net_src"/"net_dst", and "src"/"dst"
addresses, where an address is an address type, an address length in
bytes, and a pointer to that many bytes.
"dl_{src,dst}" are the link-layer source/destination; "net_{src,dst}"
are the network-layer source/destination; "{src,dst}" are the
source/destination from the highest of those two layers that we have in
the packet.
Add a port type to "packet_info" as well, specifying whether it's a TCP
or UDP port.
Don't set the address and port columns in the dissector functions; just
set the address and port members of the "packet_info" structure. Set
the columns in "fill_in_columns()"; this means that if we're showing
COL_{DEF,RES,UNRES}_SRC" or "COL_{DEF,RES,UNRES}_DST", we only generate
the string from "src" or "dst", we don't generate a string for the
link-layer address and then overwrite it with a string for the
network-layer address (generating those strings costs CPU).
Add support for "conversations", where a "conversation" is (at present)
a source and destination address and a source and destination port. (In
the future, we may support "conversations" above the transport layer,
e.g. a TFTP conversation, where the first packet goes from the client to
the TFTP server port, but the reply comes back from a different port,
and all subsequent packets go between the client address/port and the
server address/new port, or an NFS conversation, which might include
lock manager, status monitor, and mount packets, as well as NFS
packets.)
Currently, all we support is a call that takes the source and
destination address/port pairs, looks them up in a hash table, and:
if nothing is found, creates a new entry in the hash table, and
assigns it a unique 32-bit conversation ID, and returns that
conversation ID;
if an entry is found, returns its conversation ID.
Use that in the SMB and AFS code to keep track of individual SMB or AFS
conversations. We need to match up requests and replies, as, for
certain replies, the operation code for the request to which it's a
reply doesn't show up in the reply - you have to find the request with a
matching transaction ID. Transaction IDs are per-conversation, so the
hash table for requests should include a conversation ID and transaction
ID as the key.
This allows SMB and AFS decoders to handle IPv4 or IPv6 addresses
transparently (and should allow the SMB decoder to handle NetBIOS atop
other protocols as well, if the source and destination address and port
values in the "packet_info" structure are set appropriately).
In the "Follow TCP Connection" code, check to make sure that the
addresses are IPv4 addressses; ultimately, that code should be changed
to use the conversation code instead, which will let it handle IPv6
transparently.
svn path=/trunk/; revision=909
Guy Harris