This commit reduces size (from 144B to 128B on AMD64) of frame_data structure.
Part of bug 5821: Reduce per-packet memory requirements.
svn path=/trunk/; revision=45071
make Save-As/Displayed/All-Packets save not only the displayed packets but
also any other packets needed (e.g., for reassembly) to fully dissect the
displayed packets.
This works only for the "All packets" case; choosing only the Selected packet,
the Marked packets, or a range of packets would require actually storing which
packets depend on which (too much memory) or going through the packet list many
times (too slow). Also, this behavior is always the case: you can't save the
displayed packets without their dependencies (I don't see why this would be
desirable).
So far this is done for SCTP and things using the reassembly routines (TCP has
been tested).
The Win32 dialog was modified but hasn't been tested yet.
One confusing aspect of the UI is that the Displayed count in the Save-As
dialog does not match the number of displayed packets. (I tried renaming the
button "Displayed + Dependencies" but it looked too big.) The tooltip tries
to explain this and the fact that this works only in the All-Packets case;
suggestions for improvement are welcome.
Implementation details:
Dissectors (or the reassembly code) can list frames which were needed to
build the current frame's tree. If the current frame passes the display
filter then each listed frame is marked as "depended upon" (this takes up the
last free frame_data flag).
When performing a Save-As/Displayed/All-Packets then choose packets which
passed the dfilter _or_ are depended upon.
svn path=/trunk/; revision=41216
by Wiretap, to indicate whether certain fields in that structure
actually have data in them.
Use the "time stamp present" flag to omit showing time stamp information
for packets (and "packets") that don't have time stamps; don't bother
working very hard to "fake" a time stamp for data files.
Use the "interface ID present" flag to omit the interface ID for packets
that don't have an interface ID.
We don't use the "captured length, separate from packet length, present"
flag to omit the captured length; that flag might be present but equal
to the packet length, and if you want to know if a packet was cut short
by a snapshot length, comparing the values would be the way to do that.
More work is needed to have wiretap/pcapng.c properly report the flags,
e.g. reporting no time stamp being present for a Simple Packet Block.
svn path=/trunk/; revision=41185
This is POC we may want to have more efficient use of the frame data
structure etc. But this allows for work to be done on the GUI to actually add comments.
svn path=/trunk/; revision=40969
Added ability to display UTC time or UTC time with date. I liked having the
difference between UTC and local time, not just setting local=UTC.
https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=2629
svn path=/trunk/; revision=37898
sequence of frame_data structures, indexed by the frame number. Extract
the relevant bits of the capture_file data structure and move them to
the frame_data_sequence, and move the relevant code from cfile.c and
tweak it to handle frame_data_sequence structures.
Have a possibly-null pointer to a frame_data_sequence structure in the
capture_file structure; if it's null, we aren't keeping a sequence of
frame_data structures (we don't keep that sequence when we're doing
one-pass processing in TShark).
Nothing in libwireshark should care about a capture_file structure; get
rid of some unnecessary includes of cfile.h.
svn path=/trunk/; revision=36881
This lets us get rid of the per-frame_data-structure prev and next
pointers, saving memory (at least according to Activity Monitor's report
of the virtual address space size on my Snow Leopard machine, it's a
noticeable saving), and lets us look up frame_data structures by frame
number in O(log2(number of frames)) time rather than O(number of frames)
time. It seems to take more CPU time when reading in the file, but
seems to go from "finished reading in all the packets" to "displaying
the packets" faster and seems to free up the frame_data structures
faster when closing the file.
It *is* doing more copying, currently, as we now don't allocate the
frame_data structure until after the packet has passed the read filter,
so that might account for the additional CPU time.
(Oh, and, for what it's worth, on an LP64 platform, a frame_data
structure is exactly 128 bytes long. However, there's more stuff to
remove, so the power-of-2 size is not guaranteed to remain, and it's not
a power-of-2 size on an ILP32 platform.)
It also means we don't need GLib 2.10 or later for the two-pass mode in
TShark.
It also means some code in the TCP dissector that was checking
pinfo->fd->next to see if it's NULL, in order to see if this is the last
packet in the file, no longer works, but that wasn't guaranteed to work
anyway:
we might be doing a one-pass read through the capture in TShark;
we might be dissecting the frame while we're reading in the
packets for the first time in Wireshark;
we might be doing a live capture in Wireshark;
in which case packets might be prematurely considered "the last packet".
#if 0 the no-longer-working tests, pending figuring out a better way of
doing it.
svn path=/trunk/; revision=36849
encapsulation/data-link type to 16 bits, and shuffle some fields to
eliminate some unnecessary padding - the net result should be no change
in the structure size for 32 bits and a few bytes removed for 64 bits.
This allows more encapsulation types - we've just about run out of the
ones that fit in a signed 8-bit integer - and thus should fix bug 5025.
svn path=/trunk/; revision=33613
This patch is cut&paste code from gtk/main_packet_list.c:packet_list_compare()
to new function frame_data_compare()
+ it make use of new function inside packet_list_compare() and
packet_list_compare_records()
svn path=/trunk/; revision=29164
they have LF at the end of the line on UN*X and CR/LF on Windows;
hopefully this means that if a CR/LF version is checked in on Windows,
the CRs will be stripped so that they show up only when checked out on
Windows, not on UN*X.
svn path=/trunk/; revision=11400
equivalents for the epan/ directory but leave winsock2.h in inet_pton.c
and inet_ntop.c for now (can't estimate the consequences).
svn path=/trunk/; revision=5928