routines that don't return. (This requires that some files include
config.h to get WS_MSVC_NORETURN declared properly.)
svn path=/trunk/; revision=35989
- Define macros for certain CFLAGS in config.nmake iso of having defs in each makefile;
a. -DHAVE_CONFIG_H and -D_U_="" are now part of a macro named STANDARD_CFLAGS;
b. -WX has been replaced by WARNINGS_ARE_ERRORS (defined as -WX in config.nmake)
(This allows disabling "Warnings as Errors" by just changing config.nmake)
c. CVARSDLL definitions (not usage) have been removed from the various makefiles.
XXX: It appears the usage of CVARSDLL can also be removed (not yet done) since:
-DWIN32 and -DNULL=0 do not appear to be needed (any more);
-D_MT and _D_DLL are not needed since /MP causes these definitions.
d. Define a macro WARNINGS_CFLAGS with additional specific compiler (level4) warnings to be enabled.
E.G., 4295: array is too small to include a terminating null character
- config.nmake: reformat some long lines for readability;
- plugins\Makefile.nmake: clean-deps does nothing: remove it (and usage in top-level makefile);
- dissectors/Makefile.nmake: test to enable packet-rrc.obj target needs to include MSVC2010 ...
svn path=/trunk/; revision=35747
I'm reasonably sure that I introduced this bug and I apologize for the problems
with my previous patch. The problem is that I did not use all of the seen
keys, I used all except the first key, which in a case of one key is none.
The attached patch fixes the error.
svn path=/trunk/; revision=29843
its own crypt-aes.
change the integer types to glib style integers
this may/will be helpful if/when we implement our own version of
kerberos
aes decryption of dcerpc since the existing libraries can not (yet)
handle when header signing is used.
we should implement our own decryption of this for cfx+aes just as we
did for classic+arcfour
svn path=/trunk/; revision=29228
Airpdcap does not allow for more than one key to be stored for a pair of nodes.
This means that when a device associates more than once the previous keys are
lost. This is ok for the first pass as the newest key is all that is needed
but when the user tries to click on a packet, to get the tree, which used a
previous key all that is seen is the encrypted data. The attached patch stores
previous associations in a linked list and will try all known keys before
decided the packet can't be decrypted. The list of keys is garbage collected
when a new capture is started.
svn path=/trunk/; revision=28449
Although this patch successfully recognizes group keys and decrypts packets
properly using the group key, there is a limitation. If an AP is using key
rotation, clicking on individual packets in a trace may not properly decrypt a
packet encrypted with a group key. This is because the current structure used
in Wireshark only supports one active unicast and one active group key. If a
new key has been seen, but you are looking at a packet encrypted with an older
key, it will not decrypt. The summary lines, however, do show the packets
properly decrypted.
I've written up a much longer and more detailed explanation in a comment in the
code, along with a proposed idea for a solution, plus a clunky work-around in
the GUI when using the current code.
I also suspect there might still be a problem with decrypting TKIP groups keys
that are sent using WPA2 authentication. In the most common operation, if you
are using WPA2, you'll also be using AES keys. It's not a common AP
configuration to use WPA2 with TKIP. In fact, most APs don't seem to support
it. Since it is an uncommon setup, I haven't put aside the time to test this
patch against such an AP. I do have access to an AP that supports this, so
when I have the time I'll test it and if needed, will submit another patch to
handle that odd-ball condition.
From me:
Remove the decrypt element of s_rijndael_ctx (which was unused, as indicated
in the comments).
Preserve the GPL licensing text in several files (which the patch shouldn't
have removed).
Remove changes that added whitespace.
Convert C++-style comments to C-style.
Update to include recent SVN changes (e.g. renaming variables named "index").
Remove extraneous printf's.
Define DEBUG_DUMP in airpdcap_debug.h.
Comment out some instances of DEBUG_DUMP.
Change malloc/free to g_malloc/g_free.
Use g_memdup instead of allocating and copying.
Use gint16 instead of INT16 in airpdcap_rijndael.c.
Add Brian to AUTHORS.
svn path=/trunk/; revision=25879
still declared by <string.h> on some platforms (at least the way we
compile, with all sorts of non-ANSI C/non-POSIX stuff added).
svn path=/trunk/; revision=25551
- 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
and call AirPDcapInitContext() where we were previously calling
AirPDcapCleanKeys(). If we're resetting our keys, we should reset our
SA list and other associated data as well.
svn path=/trunk/; revision=24562
est. Use g_ascii_strcasecmp() and g_ascii_strncasecmp(), and supply our
own versions if they're missing from GLib (as is the case with GLib
1.x).
In the code to build the list of named fields for Diameter, don't use
g_strdown(); do our own g_ascii_-style upper-case to lower-case mapping
in the hash function and use g_ascii_strcasecmp() in the compare
function.
We do this because there is no guarantee that toupper(), tolower(), and
functions that use them will, for example, map between "I" and "i" in
all locales; in Turkish locales, for example, there are, in both
upper case and lower case, versions of "i" with and without a dot, and
the upper-case version of "i" is "I"-with-a-dot and the lower-case
version of "I" is "i"-without-a-dot. This causes strings that should
match not to match.
This finishes fixing bug 2010 - an earlier checkin prevented the crash
(as there are other ways to produce the same crash, e.g. a bogus
dictionary.xml file), but didn't fix the case-insensitive string matching.
svn path=/trunk/; revision=23623
Gerald Combs