#include winsock2.h pulls in about 90 distinct .h files
and about 140 total .h files.
Currently winsock2.h is (mostly unnecessarily) included
for each dissector via packet.h/wtap.h.
This patch removes #include winsock2.h from wtap.h and
then includes winsock2.h (or windows.h) in the
few specific places required.
With this patch, my Windows Wireshark build takes
about 30% less time.
svn path=/trunk/; revision=26535
provide a default case (returning an error) to prevent wiretap from asserting
out because we didn't set the packet encapsulation.
svn path=/trunk/; revision=26327
of adding libwsutil but somehow I missed it/got it wrong. This should solve
https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=1677#c18
Also remove the _DEPENDENCIES lines in epan and wiretap (as was done in the
top-level Makefile in rev 25930) so that automake will automatically figure out
the dependencies for us.
Since the those 2 libraries now link against libwsutil, don't make every
executable link against the library. (If this works I think we can
significantly trim the list of libraries the executables link against and just
let the libraries pull in what they need--which is, apparently, the point of
the --as-needed flag: http://www.gentoo.org/proj/en/qa/asneeded.xml ).
svn path=/trunk/; revision=26218
Fix a final eth_fopen -> ws_fopen
When configuring with --without-zlib these functions need to have some parameters tagged _U_
svn path=/trunk/; revision=26212
do *not* modify the string handed to them - they g_mallocate a new
string and return it.
Create routines that *do* ASCII-only case mapping in place, and use them
instead.
Clean up indentation.
svn path=/trunk/; revision=26131
MSC_VER_REQUIRED when we run mt.exe instead of checking for each
individual MSVC_VARIANT. This fixes the current buildbot test failures
on Windows, which resulted from a missing check for MSVC2008. This
also keeps us from having to mess with a bunch of makefiles when we add
support for new Visual C++ versions.
svn path=/trunk/; revision=26052
From me:
Instead of adding adns_config.h, place it a custom adns package in
wireshark-win32-libs. Update tools/win32-setup.sh accordingly.
Split the MSVC2008EE variant into MSVC2008 and MSVC2008EE, similar to
MSVC2005 and MSVC2005EE. We have to worry about vcredist_x86.exe in
both cases.
Add Pascal to AUTHORS.
Update the Developer's Guide.
svn path=/trunk/; revision=25921
ERF files can contain records of type TYPE_PAD. These records are not related
to captured packets, have a zero timestamp value and no associated packet data.
Normally TYPE_PAD records are stripped out during capture, but in rare cases
unstripped files may exist.
Previously wiretap/erf.c generated an 'unknown record encapsulation' error when
encountering TYPE_PAD records.
With this patch Wireshark skips over any TYPE_PAD records within ERF traces
files without reporting an error. TYPE_PAD records are not counted, displayed
or decoded.
svn path=/trunk/; revision=25733
any case, the detailed error string is supposed to be g_malloced....)
Fix some "snoop" to be "btsnoop", and note that this is Symbian btsnoop,
not regular snoop.
svn path=/trunk/; revision=25580
are expected to return a g_mallocated error string; that's why they fill
in a gchar **, not a const gchar **. g_strdup() the argument to
KERROR(), so it's g_mallocated.
svn path=/trunk/; revision=25398
the types of read and seek_read routines in Wiretap are supposed to be,
and get rid of the casts of pointers to those functions (type problems
should be fixed, not papered over with casts, whenver possible).
svn path=/trunk/; revision=25393
argument, as
1) it doesn't modify the string that argument points to
and
2) it's a buffer of "char".
Use g_ascii_xdigit_value() and put the values of the two bytes together
ourselves; strtoul() is a bit of overkill for two-hex-digit pairs.
While we're at it, check for invalid hex digits, and for bytes where
only one hex digit is present.
svn path=/trunk/; revision=25392
since rev 17756, meant that attempts to read iSeries files would fail in the
"Make sure it [pkt_encap] is not WTAP_ENCAP_PER_PACKET" assertion in
wtap_read().
Also set file_encap to WTAP_ENCAP_ETHERNET (instead of WTAP_ENCAP_PER_PACKET)
since it seems that all the packets in iSeries files are Ethernet (or at least
this module currently only supports Ethernet).
svn path=/trunk/; revision=25388
I was counting the length without the terminating NULL, and didn't want one to be written to the string and copied into the file.
svn path=/trunk/; revision=25383
libwireshark (and the plugins using those functions) do not depend on
wiretap on Windows.
While doing that, rename the eth_* functions to ws_*.
svn path=/trunk/; revision=25354
To prevent Windows compiler errors when using flex 2.5.35.
Fixes "missing unistd.h" and yywrap "mismatched parameter" warnings
[Upcoming Part 3: ignore 'signed /unsigned mismatch' errors]
svn path=/trunk/; revision=25173
This extends the EyeSDN wiretap module to be able to support:
- DSS1/Q.931
- PPP
- LAPB/X.25
- ATM raw cells
- SS7 MTP2
svn path=/trunk/; revision=25123
- Change ugly GLIB version checking statements to GLIB_CHECK_VERSION
- Remove ws_strsplit files because we no longer need to borrow GLIB2's
g_strsplit code for the no longer supported GLIB1 builds
svn path=/trunk/; revision=24829
This plugin implements a dissector for Infiniband. It is released
under the GPL v2.
Rather than using say libpcap to capture raw (unframed) IP packets
from near the top of an IPoIB stack, this plugin dissects link level
Infiniband frames.
Infiniband trace files can be read from Endace ERF format trace
files, or from libpcap DLT_ERF files containing ERF TYPE_INFINIBAND
records. There is currently no native DLT_INFINIBAND in libpcap.
Each record contains a hardware timestamp, capture metadata such as
port Id, and a complete link level Infiniband frame starting from
the Local Route Header.
svn path=/trunk/; revision=24628
This patch adds some new ENCAP and FILE types for wiretap. It also adds new
entries to pcap_to_wtap_map[] to provide a mapping of the new types to some
pcap DLTs.
svn path=/trunk/; revision=24622
point, so we don't have issues with numbers not being exactly
representable; that makes it less likely that the change described below
will change a time stamp if it's not fixing the time stamp (i.e., if
time_day is 0).
The Sniffer manual "Expert Sniffer(R) Network Analyzer Operations,
Release 5.50" says that a frame2_rec has a time stamp with an 8-bit
time_high field and an 8-bit time_day field. Interpreting the time
stamp that way fixes the time stamps in at least some captures; see, for
example, bug 2251.
Fix/update some comments (for example, the Sniffer documentation is no
longer at that URL).
svn path=/trunk/; revision=24296
Added support for Symbian OS btsnoop.
The bluetooth HCI layer in Symbian OS can be configured to log all packets to a
file. The log format, "btsnoop" is based on the RFC1761 "snoop" format - but
differences in the header make it incompatible.
The btsnoop format supports logging of these formats:
"H1" (raw HCI packets without framing)
"H4" (HCI UART packets including packet type header)
"H5" (HCI 3 wire UART packets including framing)
"BCSP" (HCI bluecore serial protocol including framing)
"H1" and "H4" are section numbers in the original v1 bluetooth specifications,
but still used colloquially - wireshark's existing support for Linux bluez HCI
logs uses the "H4" name.
In practice, the "H1" format is used for H5,BCSP and USB HCI logs, as the HCI
packet logs are mainly useful for debugging higher layers, bluetooth profiles
and bluetooth applications.
From me:
Deleted some unused prototypes.
Mark an unused parameter.
svn path=/trunk/; revision=24263
otherwise use as the per-packet encapsulation.
The close routine does nothing; get rid of it - you don't *need* a close
routine (by default, the subtype_close pointer is null, which means that
nothing per-file-type is done when the file is closed).
Make the code to handle the length fields in the ERF header common,
rather than copying it to each group of record types.
svn path=/trunk/; revision=24053
fix http://bugs.wireshark.org/bugzilla/show_bug.cgi?id=1727 (pppd format file
incorrectly detected as being an ERF file) by:
The file_seek() call has been replaced by a call to file_read(), so, when the
end of the file is reached and the current record is truncated, we have got an
error.
This solves the problem of bad file format detection.
Additionaly, the ERF heuristic has been improved.
svn path=/trunk/; revision=24051
1/ patches to support the libpcap/SITA format 'WTAP_ENCAP_SITA'.
2/ patches to the LAPB dissector to accept MLP (Multi-link protocol)
(although MLP dissection has _not_ been added (yet)).
3/ New protocol dissectors for:
a) SITA's WAN layer 0 status header,
b) An airline protocol ALC,
c) An airline (and other industry) protocol UTS.
These patches are submitted as a set since the new protocol dissectors are not
useful without the libpcap/SITA related changes, and there is no point in
having those changes without the additional dissectors.
This fixes bug/enhancement 2016.
svn path=/trunk/; revision=23885
Error message when capturing too short WTAP_ENCAP_USB_LINUX type packets
contains a copy-paste typo.
From me:
Fix some addresses in AUTHORS.
svn path=/trunk/; revision=23882
no-longer-needed wiretap/wtap-capture.h.
Clean up wiretap/libwiretap.vcproj (note: this isn't going to scale, if
it has to contain a list of all the files, as most committers will be
editing only Makefile.common files).
svn path=/trunk/; revision=23803
Copy the MSVC++-version-checking stuff from it into the top-level
config.h.win32, and try to set up config.nmake so that MSC_VER_REQUIRED
is defined in all Makefiles that include config.nmake.
svn path=/trunk/; revision=23802
the top-level directory.
Don't build config.h in the wiretap directory - it now uses config.h
from the top-level directory. Get rid of the template config.h.win32 in
that directory.
Get rid of files that are no longer in the wiretap directory or are no
longer constructed in that directory.
svn path=/trunk/; revision=23797
in Wireshark when reading CommView files. Also write out these values when
coming from a file format with encapsulation type 802.11 with radio.
svn path=/trunk/; revision=23617
used by the mpeg-audio dissector: instead keep the data inside the wiretap
module and add accessor functions. I think this should fix
http://bugs.wireshark.org/bugzilla/show_bug.cgi?id=1677 and anyway it's
cleaner.
svn path=/trunk/; revision=23612
HDLC" to try to distinguish Cisco HDLC from PPP packets;
Address-and-Control-Field-Compression can be negotiated on, so there's
no guarantee that PPP frames begin with 0xff 0x03. Fixes bug 2005.
svn path=/trunk/; revision=23535
When trying to open a pcap file with the new pseudo-header/DLT (using SVN
version, changelist 23283) I get the error message:
"libpcap: ERF file has a 13-byte packet, too small to have even an ERF
pseudo-header".
After reviewing Paolo's patch I found that there are 2 places with missing
breaks in switch case structures.
svn path=/trunk/; revision=23298
using modf() and thus not requiring libm. In addition, adding -lm to
the dependencies upsets the build on at least some platforms because
"-lm" isn't a pathname.
svn path=/trunk/; revision=23210
http://bugs.wireshark.org/bugzilla/show_bug.cgi?id=1751
The patch adds support to wiretap for a new libpcap DLT for bluetooth captures.
This DLT carries the direction information, which now can be displayed
correctly.
The hci H4 dissector is updated to handle also the newly introduced wtap encap.
svn path=/trunk/; revision=23208
This is a replacement of the existing decoding of ERF files (Extensible Record
Format from Endace).
For the decoding of the ERF files, according to the "type of record" given in
the ERF header, several decoders can be used. Up to now, the decoder is
determined according to an environment variable, or with a kind of heuristic.
And, all the treatment is done during the file extraction.
The new architecture, will separate the ERF file decoding, and the ERF record
decoding. The ERF records will be decoded with a specific dissector. This
dissector can be configured with options, to replace the environment variable.
http://bugs.wireshark.org/bugzilla/show_bug.cgi?id=1839
svn path=/trunk/; revision=23092
if there are packets with different encapsulationtype in the file.
Otherwise use the encapsulationtype of the packets in the file.
This makes it possible to save the imported data as libpcap file
(or any other format that does not support per-packet encapsulation).
svn path=/trunk/; revision=23031
Attached is a small patch that correct an issue with reading certain IBM
iSeries Comms traces.
Traces where data has been dropped for whatever reason now have the
packet number suffixed with an asterix "*", this causes the current
iSeries wiretap routine to report a "bad" header. The attached patch
simply scans the packet number field and removes any "* characters prior
to scanning, the fact that data may be missing is more than adequately
reported later by current wireshark packet processing.
Regards .. Martin
svn path=/trunk/; revision=23000
tshark can read a HP-UX nettl IP packet dump (written with 'nettl -traceon all
-entity ns_ls_ip -file dump'), but cannot convert it to a pcap raw IP packet
dump, with 'tshark -r dump.nettl -w dump.pcap'. A single-line patch to
wiretap/libpcap.c makes it possible to do this.
The input file uses encapsulation type WTAP_ENCAP_NETTL_RAW_IP.
svn path=/trunk/; revision=22849
- It accepts the "/" character in interface names
- It accepts EOF as delimiter for the last packet (when there is no more emptyline)
svn path=/trunk/; revision=22765
Use G_GINT64_CONSTANT() to make a 64-bit integral constant; not all
compilers we use support LL as a suffix for that (MSVC++ 6, for one).
svn path=/trunk/; revision=22580
This patch adds support for the Juniper NetScreen snoop output format.
It takes a text-dump op the captured packets and parses the headers
and hex-data. Since the snoop files on a Junpiper NetScreen can be saved
to a tftp-server, this patch makes it quite easy to use the snoop
function of the Juniper NetScreen firewalls.
/* XXX TODO:
*
* o Create a wiki-page with instruction on how to make tracefiles
* on Juniper NetScreen devices. Also put a few examples up
* on the wiki (Done: wiki-page added 2007-08-03)
*
* o Use the interface names to properly detect the encapsulation
* type (ie adsl packets are now not properly dissected)
* (Done: adsl packets are now correctly seen as PPP, 2007-08-03)
*
* o Pass the interface names and the traffic direction to either
* the frame-structure, a pseudo-header or use PPI. This needs
* to be discussed on the dev-list first
* (Posted a message to wireshark-dev abou this 2007-08-03)
*
*/
svn path=/trunk/; revision=22533
1) "-e" isn't supported by good old /bin/sh, so we use "-r"
instead;
2) "The algorithm for determining the precedence of the
operators and the return value that will be generated is
based on the number of arguments presented to test", so we
explicitly parenthesize.
svn path=/trunk/; revision=22448
such as the fact that Flex strips all but the last component of the "-o"
argument, and that it doesn't generate a header file to declare routines
the generated lexical analyzer defines. Use that script when building
lexical analyzers, and, for each lexical analyzer, include the generated
header file in the generated analyzer.
svn path=/trunk/; revision=22446
Makefile.nmake files; currently, it has the (F)lex-to-C rule and a
.SUFFIXES pseudo-rule to add .l to the list of suffixes. Have
Makefile.nmake files with .l.c rules include Makefile.nmake.inc to get
that rule.
The names Makefile.am.inc and Makefile.nmake.inc aren't necessarily the
right names for the files in question.
Use $(PACKAGE) in the Mate plugin's Makefile, rather than "mate".
svn path=/trunk/; revision=22437
Makefile.am files; currently, it has the (F)lex-to-C rule. Have
Makefile.am files with .l.c rules include Makefile.am.inc to get that
rule.
svn path=/trunk/; revision=22436
Move the %options to the beginning if they weren't already there, and
put them in the same order in all files.
Add "prefix=" options to .l files that don't already have them, so we
don't have to pass a "-P" option.
Add "never-interactive" and "noyywrap" options to our lexical analyzers,
to remove extra isatty() checks and to eliminate the need for yywrap()
from the Flex library.
Get rid of %option nostdinit - that's the default.
Add .l.c: rules to Makefile.am files, replacing the rules for specific
.l files. Have those rules all check that $(LEX) is set.
Update the address for the FSF.
svn path=/trunk/; revision=22424
a source release tarball without having Flex (think of a source release
tarball being as much a platform-independent distribution format for
people *not* interested in development, and who are on platforms for
which there aren't binary packages, as a way of getting the source to do
development). Don't check Flex's capabilities in the configure script
(handling reentrant scanners would have to be done differently).
svn path=/trunk/; revision=22414
Its argument, however, needs to be cast to "guchar", so that if the
high-order bit is set, it doesn't get sign-extended.
svn path=/trunk/; revision=22303
The encap_table_base in wcap.c is missing an entry.
This causes e.g. "dumpcap -i usb3 -L" to output:
Data link types (use option -y to set):
USB_LINUX
(MPEG)
svn path=/trunk/; revision=22292
The code for reading ERF files has not been significantly
updated since 2004. This patch brings it up to date with a
number of changes.
1) Increase number of decodable ERF types from 7 to 12. This
covers newer DAG card models and firmware updates.
2) Fix timestamp conversion. Was calculating only microsecond
precision, now displaying with nanosecond resolution. Hardware
precision is 7.5 to 30 ns depending on model.
3) Allow the user to specify HDLC encapsulation as 'chdlc',
'ppp_serial', 'frelay' or 'mtp2'. This is needed because the
ERF HDLC capture formats do not include information on what
protocol is used at the next level. This is currently done via
an environment variable 'ERF_HDLC_ENCAP' and is analagous to the
existing 'ERF_ATM_ENCAP' variable.
If the user does not specify an HDLC encapsulation it tries to
guess, and falls back to MTP2 for backwards compatibility with
Florent's existing behaviour.
I know environment variables are ugly, suggestions are welcome.
4) When reading HDLC captures as MTP2, use
WTAP_ENCAP_MTP2_WITH_PHDR rather than WTAP_ENCAP_MTP2. This
allows us to put the 'Multi-Channel ERF' record 'channel
number' field into the MTP2 pseudo header > 'link_number'
field. This is then displayed in Frame information, and can
be filtered on. (Would be nice if it could be made a display
column?)
Because the ERF record does not specify whether Annex A is used
or not, we pass MTP2_ANNEX_A_USED_UNKNOWN and allow the existing
user preference to decide.
Move the MTP2_ANNEX_A_ definitions into Wiretap, make the annex_a_used
field a guint8, and change MTP2_ANNEX_A_USED_UNKNOWN to 2 so it fits in
a guint8. (This means that if you can save an ERF MTP2 file as a
libpcap file, the pseudo-header will have MTP2_ANNEX_A_USED_UNKNOWN in
it.)
svn path=/trunk/; revision=22067
network type; there's no "presumably" about it.
Suggest that "realtick" might have the right time stamp in other cases
(if not, a comment should explicitly indicate that, so that in all cases
where we either know that realtick is wrong or have a lot of evidence to
show that it's right, we note that fact).
svn path=/trunk/; revision=21996
Fix compilation failures when building wireshark-0.99.6-SVN-21916 on an
x86_64-unknown-linux-gnu target with gcc version 4.1.2 20070403 (Red Hat
4.1.2-8).
The failures fall into two categories:
(1) Casts between pointers and 32-bit integers without an intermediary cast
via 'long' or 'unsigned long'. This results in a compiler warning complaining
about casts between a pointer and an integer of a different size.
(2) Passing values to "%lld" or similar printf-style format options that the
compiler thinks are a different size. Such values need to be cast to 'long
long' or 'unsigned long long'.
svn path=/trunk/; revision=21975
possibly-unaligned pointers, and turn on -Wcast-align so at least some
future code that does that will fail to compile.
svn path=/trunk/; revision=21968
what the complete set of warnings we should either try to fix or, for
cases where it can't be fixed, turn off or or avoid -Werror for. I'll
revert this change as soon as a complete set of buildbot builds start
with it.
svn path=/trunk/; revision=21917
libraries. A single library is generated with the lex code without the barrier
"stop on warning". An other library is generated from the remaining source
files with the "stop on warning" barrier.
svn path=/trunk/; revision=21817
Since wiretap.h is not autogenerated, just make wtap-plugins.h include
the top level config.h to pull in the defines for HAVE_DIRENT which was
checked at top level
svn path=/trunk/; revision=21707
So far I've done only regression testing (the new functionality and what's in wtap-plugins.c has not yet being tested).
it is a first step in the way to have lua opening files.
svn path=/trunk/; revision=21686
--enable-warnings-as-errors (if any tests enabled by
--enable-extra-gcc-checks are safe to treat as errors, they're safe to
turn on by default).
svn path=/trunk/; revision=21515
on some platforms due to unfixable problems (e.g., crappy vendor
headers), we can move them back to the "extra" list.
Put those warnings in the order in which they appear in the GCC man page
on my machine.
If we turn on -pedantic, try turning on -Wno-long-long as well, so that
it's not *so* pedantic that it rejects the 64-bit integral data types
that we explicitly require.
svn path=/trunk/; revision=21514
Stephen Fisher