"Follow Stream" functionality assumes that all data in a single packet
belongs to the same stream. That is not true for HTTP2 and QUIC, where
we end up having data from unrelated streams.
Filter out the unwanted data directly in the protocol dissector code with
a custom `tap_handler` (as TCP already does).
Close#16093
The secs member of nstime_t is a time_t, which is difficult to print in
a way that's compatible across platforms. Convert our time to floating
point and print that value instead. Fixes
../ui/cli/tap-follow.c:304:63: error: format specifies type 'unsigned long long' but the argument has type 'time_t' (aka 'long') [-Werror,-Wformat]
printf(" timestamp: %" G_GINT64_MODIFIER "u.%09d\n", follow_record->abs_ts.secs, follow_record->abs_ts.nsecs);
~~~~~~~~~~~~~~~~~~~~~~~ ^~~~~~~~~~~~~~~~~~~~~~~~~~
%ld
on macOS.
Modify YAML output format so it includes information about peers and
absolute timestamps for each packet.
This also adds yaml output to tshark: -z follow,tcp,yaml,X
The QUIC transport protocol provides a stream, similar to HTTP/2. Make
it possible to look at the stream contents. This can be helpful while
HTTP/3 support is not yet complete.
Known issues that will be addressed in the future:
- If a single packet contains multiple streams, then Follow QUIC Stream
will wrongly include data from streams other than the selected one.
This is tracked by bug 16093 and affects HTTP/2 as well.
- The Substream index menu does not properly filter for available
stream numbers. If a non-existing stream is selected, then changing
to another (potentially valid) index results in the "Capture file
invalid." error. As workaround, clear the display filter first.
- Follow Stream always selects Stream ID 0 instead of the first or
currently selected stream field in a packet. Users should manually
update the stream index as needed.
Change-Id: I5866be380d58c96f0a71a29abdbd1be20ae3534a
Ping-Bug: 13881
Reviewed-on: https://code.wireshark.org/review/34694
Petri-Dish: Peter Wu <peter@lekensteyn.nl>
Tested-by: Petri Dish Buildbot
Reviewed-by: Alexis La Goutte <alexis.lagoutte@gmail.com>
Reviewed-by: Anders Broman <a.broman58@gmail.com>
The HTTP/2 protocol multiplexes a single TCP connection into multiple
independent streams. The Follow TCP output can interleave multiple
HTTP/2 streams, making it harder to analyze a single HTTP/2 stream.
Add the ability to select HTTP/2 Streams within a TCP stream.
Internally, the HTTP/2 dissector now stores the known Stream IDs in a
set for every TCP session which allows an amortized O(n) lookup time for
the previous/next/max Stream ID.
[Peter: make the dissector responsible for clamping the HTTP/2 Stream ID
instead of the Qt code, that should permit future optimizations.]
Change-Id: I5d78f29904ae8f227ae36e1a883155c0ed719200
Reviewed-on: https://code.wireshark.org/review/32221
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Petri-Dish: Peter Wu <peter@lekensteyn.nl>
Tested-by: Petri Dish Buildbot
Reviewed-by: Alexander Gryanko <xpahos@gmail.com>
Reviewed-by: Alexis La Goutte <alexis.lagoutte@gmail.com>
Change all wireshark.org URLs to use https.
Fix some broken links while we're at it.
Change-Id: I161bf8eeca43b8027605acea666032da86f5ea1c
Reviewed-on: https://code.wireshark.org/review/34089
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Register the follow_free as a finish callback.
Call the tap's finish callback in free_tap_listener which is called from
both remove_tap_listener and tap_cleanup.
Describe the finish callback in README.tapping
Change-Id: Ie79ce8f39b0ca6d84810d2e3307dcff38b18d3ac
Reviewed-on: https://code.wireshark.org/review/29309
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Petri-Dish: Peter Wu <peter@lekensteyn.nl>
Tested-by: Petri Dish Buildbot
Reviewed-by: Michael Mann <mmann78@netscape.net>
Reverse the payload chunks list to achieve a running time of O(n) rather
than O(n²) for insertion of all chunks. Executing a RelWithDebInfo+ASAN
build with `tshark -r chargen-session.pcapng.gz -qz follow,tcp,hex,0`
previously took 11m5s to complete, but now finishes in 16 seconds.
Tested using a capture file with 152k TCP packets (from bug 11777).
Backport note: must update ui/gtk/follow_stream.c too.
Change-Id: Icf70d45f33d4399e53209fb6199d3809608c8d99
Reviewed-on: https://code.wireshark.org/review/28595
Petri-Dish: Peter Wu <peter@lekensteyn.nl>
Tested-by: Petri Dish Buildbot
Reviewed-by: Anders Broman <a.broman58@gmail.com>
Move */ to a separate line below the SPDX identifier.
Change-Id: Id1032215449cfccae0933147b45e04b65e0b727f
Reviewed-on: https://code.wireshark.org/review/27211
Reviewed-by: Anders Broman <a.broman58@gmail.com>
If we're constructing the string, assign the pointer to it to a
non-const pointer variable, set the stat_tap_ui cli_string member to
that variable, and then use the variable to free it when we're done.
Don't cast away constness if we don't have to.
Change-Id: If3b24dbf1c910e1e6eceb76f2f6a7ae3898315f9
Reviewed-on: https://code.wireshark.org/review/25952
Reviewed-by: Guy Harris <guy@alum.mit.edu>
The first is deprecated, as per https://spdx.org/licenses/.
Change-Id: I8e21e1d32d09b8b94b93a2dc9fbdde5ffeba6bed
Reviewed-on: https://code.wireshark.org/review/25661
Petri-Dish: Anders Broman <a.broman58@gmail.com>
Petri-Dish: Dario Lombardo <lomato@gmail.com>
Reviewed-by: Anders Broman <a.broman58@gmail.com>
We have WS_INET_ADDRSTRLEN and WS_INET6_ADDRSTRLEN; use them.
Change-Id: Idade0da9fae70d891901acd787b06d21e2ddbc5f
Reviewed-on: https://code.wireshark.org/review/24156
Reviewed-by: Guy Harris <guy@alum.mit.edu>
That allows a parallel typedef of ws_in4_addr for guint32.
Change-Id: I03b230247065e0e3840eb87635315a8e523ef562
Reviewed-on: https://code.wireshark.org/review/24073
Reviewed-by: Guy Harris <guy@alum.mit.edu>
That way, the compiler ensures proper alignment. In practice, the
alignment was probably proper anyway, but this makes sure.
Change-Id: I5ddc028c97d6961692a297cac17236206b61169d
Reviewed-on: https://code.wireshark.org/review/24061
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Change-Id: If7a28ba305bed9f63f200f8a8a27a796f2df9c07
Reviewed-on: https://code.wireshark.org/review/20015
Petri-Dish: Michael Mann <mmann78@netscape.net>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Michael Mann <mmann78@netscape.net>
Use the model from the 2.0 branch and earlier that only "tapped" the
follow data in a single location. This fixes duplicate data for
reassembled data and handles out-of-order packets.
Bug: 12855
Change-Id: I5268f13e3c08e9271acf026b859de693ad794c94
Reviewed-on: https://code.wireshark.org/review/18368
Petri-Dish: Pascal Quantin <pascal.quantin@gmail.com>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Pascal Quantin <pascal.quantin@gmail.com>
Chunk 0 is not allowed, so it's discarded when outputting a stream.
Ping-Bug: 12855
Change-Id: Ia7bbb0e222b6b4e714f8bfa245e910c151d82905
Reviewed-on: https://code.wireshark.org/review/18034
Petri-Dish: Michael Mann <mmann78@netscape.net>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Michael Mann <mmann78@netscape.net>
Change-Id: I5669e2442582f899643fae4a9f86ab6d505dde07
Reviewed-on: https://code.wireshark.org/review/16505
Petri-Dish: Michael Mann <mmann78@netscape.net>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Michael Mann <mmann78@netscape.net>
Some needed to check return value, others were converted to use strtoul.
Change-Id: I55aae216f95362b67e006f6e682abbd5ae2c8dcc
Reviewed-on: https://code.wireshark.org/review/16502
Petri-Dish: Michael Mann <mmann78@netscape.net>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Michael Mann <mmann78@netscape.net>
Create a "registration" system for Follow functionality so most of the work can be abstracted into a dissector and GUI can just be responsible for "display".
This also removes the global variables in follow.c to open up multithreading possibilities.
TCP, UDP and HTTP all have the same "tap interface" for Follow functionality (passing a tvb with byte data to "follow"). SSL still has it's own behavior, so Follow structures have to take that into account.
TShark through the Follow registration now has support for HTTP.
The only thing possibly missing is dynamic menu generation to further reduce explicit knowledge of Follow "type" (and rely on registration)
Bug: 11988
Change-Id: I559d9ee1312406ad0986d4dce9fa67ea2103b339
Reviewed-on: https://code.wireshark.org/review/13161
Reviewed-by: Michael Mann <mmann78@netscape.net>
No need for the GTK dependency on dissectors/packet-ipv6.h.
Add the stream_addr typedef in follow.h to make some code simpler.
Change-Id: I1cf906f58734a90263141362f2da33a140f93533
Reviewed-on: https://code.wireshark.org/review/13063
Reviewed-by: João Valverde <j@v6e.pt>
Petri-Dish: Michael Mann <mmann78@netscape.net>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Michael Mann <mmann78@netscape.net>
[KISS - Keep It Simple, Stupid]
Convert the Follow TCP functionality to use a tap from the TCP dissector that passes the tvb of the payload. This makes things A LOT simpler, but relies on the TCP dissector to make all decisions.
The "tap" logic passes tvb data
1. Before calls to process_tcp_payload
2. Before hf_tcp_segment_data fields (that aren't retransmissions or otherwise handled)
Follow up patches will be necessary to clean up all of the supporting "follow" functionality that is now useless.
Bug: 6925
Bug: 9780
Change-Id: I4e7f5d453519be839de39a109bafa899b9987139
Reviewed-on: https://code.wireshark.org/review/13038
Reviewed-by: Michael Mann <mmann78@netscape.net>
Change-Id: I3817d12e473b67e26159e1562a08169e91f51d46
Reviewed-on: https://code.wireshark.org/review/13019
Petri-Dish: Michael Mann <mmann78@netscape.net>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Michael Mann <mmann78@netscape.net>
Be more consistent about using the ws_ routines, as we suggest in
README.developer.
In C++ on UN*X, define ws_close as ::close rather than close, so that it
works even in classes with methods or members named "close".
Change-Id: Ide2652229e6b6b4624cbddae0e909a4ea1efa591
Reviewed-on: https://code.wireshark.org/review/11637
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Have wsutil/file_util.h include them on UN*X, just as it includes io.h
on Windows, so we can have a rule of "if you do file operations, include
<wsutil/file_util.h> and use the routines in it".
Remove includes of unistd.h, fcntl.h, and sys/stat.h that aren't
necessary (whether because of the addition of them to wsutil/file_util.h
or because they weren't needed in the first place).
Change-Id: Ie241dd74deff284e39a5f690a297dbb6e1dc485f
Reviewed-on: https://code.wireshark.org/review/11619
Petri-Dish: Guy Harris <guy@alum.mit.edu>
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Replace CMP_ADDRESS, COPY_ADDRESS, et al with their lower-case
equivalents in the ui directory.
Change-Id: I10e95e66c8da5b880133452ebc484c53046e87ba
Reviewed-on: https://code.wireshark.org/review/11199
Reviewed-by: Michael Mann <mmann78@netscape.net>
Fix some (mostly spurious) checkapi warnings by renaming the
offending variables.
Change-Id: I7a43ac89f5ed35053a6526fa838fbad67669a49a
Reviewed-on: https://code.wireshark.org/review/10655
Petri-Dish: Graham Bloice <graham.bloice@trihedral.com>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Graham Bloice <graham.bloice@trihedral.com>
Found by MSVC2013 Code Analysis
Change-Id: I58063946dd558e98308c87b36eeac0ddbe1a6e79
Reviewed-on: https://code.wireshark.org/review/7045
Reviewed-by: Bill Meier <wmeier@newsguy.com>
-z "follow,udp" tshark cli command now supports a stream index
It is now possible to select the UDP stream displayed in Qt GUI (like for TCP)
Change-Id: Ia367f36ea4f60db0fddb997a7e0903c09e172f2d
Reviewed-on: https://code.wireshark.org/review/6083
Reviewed-by: Pascal Quantin <pascal.quantin@gmail.com>
The intent is to handle more than just command-line arguments; reflect that.
Change-Id: Ia10efda85a9d11c6579d1bec6f789cee30d9e825
Reviewed-on: https://code.wireshark.org/review/5304
Reviewed-by: Guy Harris <guy@alum.mit.edu>
The old routine had stat_ in the name, as the expectation was that they
were for statistics taps; that's still the expectation, so have stat_ in
the data structure and routine names.
Change-Id: Ic98d011012b8641173d41fa0ec4f4e625614370a
Reviewed-on: https://code.wireshark.org/review/5303
Reviewed-by: Guy Harris <guy@alum.mit.edu>
The intent here is to centralize more UI information so that we can move
more tap UI stuff to common code. This is a beginning.
Change-Id: Ic35ac0c01bc7b942aab88177db4065847a5e6c30
Reviewed-on: https://code.wireshark.org/review/5301
Reviewed-by: Guy Harris <guy@alum.mit.edu>
With tshark stats are being configured before the file gets loaded and the number of TCP streams are computed
Bug: 9541
Change-Id: I42c2891124f1781b05967d5f071ad40df2d6d9f5
Reviewed-on: https://code.wireshark.org/review/1598
Reviewed-by: Evan Huus <eapache@gmail.com>
Reviewed-by: Pascal Quantin <pascal.quantin@gmail.com>
(Using sed : sed -i '/^ \* \$Id\$/,+1 d')
Fix manually some typo (in export_object_dicom.c and crc16-plain.c)
Change-Id: I4c1ae68d1c4afeace8cb195b53c715cf9e1227a8
Reviewed-on: https://code.wireshark.org/review/497
Reviewed-by: Anders Broman <a.broman58@gmail.com>
João Valverde