nghttp2ver.h is a local header, so use "" not <>
See commit 4a182d48b1
Change-Id: I49210906290210fde8f496996e0e5ae69d697338
Reviewed-on: https://code.wireshark.org/review/3349
Reviewed-by: Evan Huus <eapache@gmail.com>
This patch avoids the following warning with `clang -fsanitize=undefined`:
epan/dissectors/packet-lbtrm.c:1519:36: runtime error: member access within null pointer of type 'lbtrm_hdr_t'
Change-Id: I973caf92547f5d16c0de20908b2b3fbc09227df0
Reviewed-on: https://code.wireshark.org/review/3313
Reviewed-by: Evan Huus <eapache@gmail.com>
The shift `(gint32)0xFF << 24` invokes undefined behavior as it may not
fit in a signed integer. Fix this by explicitly casting 0xFF as
unsigned. Caught by `clang -fsanitize=undefined`.
While at it, convert to tvb_captured_length and add modelines.
Change-Id: I241ff8ed91815369ec0c19719750cee4b6b12343
Reviewed-on: https://code.wireshark.org/review/3311
Reviewed-by: Evan Huus <eapache@gmail.com>
On the 32-bit OS X buildbot, automake is old enough that autotools
builds are verbose, printing the command line for each compile step; if
we cause CMake to produce verbose makefiles, they will also print the
command line, so we can see what differences there are.
This means the other debugging output we added isn't necessary; remove
it.
Sort the extra warning options in configure.ac based on whether they're
for C and C++ or just C, just as is the case in CMake, to make it easier
to keep the lists in sync; add comments to configure.ac to mirror the
ones in CMakeLists.txt.
Get rid of -Wno-deprecated-declarations; the CMake build doesn't use it,
and it doesn't appear to be necessary - if we ever have problems with OS
X complaining about using OpenSSL, for example, we should do as tcpdump
does and wrap uses of the "deprecated"-but-cross-platform APIs in
the appropriate pragmas.
Change-Id: I0b8b45f00481ac2b67da5c7dd028a029b130dcc4
Reviewed-on: https://code.wireshark.org/review/3377
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Presumably that was added for tap purposes, but packet-scope is much simpler and
less dangerous. Noticed while investigating the scan-build issues with
stack-local variables being pointed to by globals.
Change-Id: I851d756b103df71079b656e624f7472354c15862
Reviewed-on: https://code.wireshark.org/review/3290
Reviewed-by: Alexis La Goutte <alexis.lagoutte@gmail.com>
Petri-Dish: Alexis La Goutte <alexis.lagoutte@gmail.com>
Petri-Dish: Anders Broman <a.broman58@gmail.com>
Reviewed-by: Anders Broman <a.broman58@gmail.com>
If the tvb contained too few data,
ssl_dissect_hnd_hello_ext_session_ticket would then allocate
session_ticket, but not initialize the contents. Fix this by adding a
check for the TVB length.
The same is done for ssl_dissect_hnd_new_ses_ticket. That might, or
might not, be necessary as proto_tree_add_item() is called with the
range. When tree is NULL, ssl is usually NULL too. For clarity (and to
avoid surprises in the future), add it anyway.
Bug: 10330
Change-Id: I469e97542542aaef4cbd660086bedf92ba1c0b6e
Reviewed-on: https://code.wireshark.org/review/3309
Reviewed-by: Evan Huus <eapache@gmail.com>
Reviewed-by: Anders Broman <a.broman58@gmail.com>
Some of the allocators are close enough that a little bit of system load at the
time of the test was enough to throw them off, and it was causing unnecessary
buildbot failures. Benchmarking on real-world cases at the time of
implementation (as well as reliable "average" results on the buildbots) has left
me confident that block_fast is, in fact, faster than block, which is faster
than simple/strict.
Additionally, the allocators are effectively "done" and unlikely to change, so
the possibility of uncaught performance regressions is low.
Change-Id: I144ea4cd9d6000edc4075ed682ce6a4aeaf5e28a
Reviewed-on: https://code.wireshark.org/review/3356
Reviewed-by: Anders Broman <a.broman58@gmail.com>
Remove our tap listeners when the capture file closes. This lets us view
conversation data for multiple capture files.
Change-Id: Ifaa1d2c9d3cd10b1d282aa0070ee2edacc7ce3b0
Reviewed-on: https://code.wireshark.org/review/3376
Reviewed-by: Gerald Combs <gerald@wireshark.org>
Hopefully this will help figure out whether we're using different
compilers for the autotools and CMake builds.
Change-Id: If6d2ee93cd14570c53723140dcedf347600cbe34
Reviewed-on: https://code.wireshark.org/review/3375
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Also, comment out the addition of -Wno-deprecated-declarations, as the
CMake build doesn't use it.
If this succeeds in making the autotools and CMake builds compile C code
with the same -W and -f flags in the same order, that should eliminate
at least one reason why the CMake build produces compiler warnings that
the autotools build doesn't.
Change-Id: I0eaeae91b8e3c39c53f502daef86c2d103c8bdab
Reviewed-on: https://code.wireshark.org/review/3374
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Now that we check each flag with all previously-accepted flags, this
catches the case where -fwrapv causes -fno-strict-overflow to produce a
warning when tested.
This way, what we do matches what the autotools do.
Change-Id: I68ce281dc33a32b591e3137db626f2749c4a0708
Reviewed-on: https://code.wireshark.org/review/3370
Reviewed-by: Guy Harris <guy@alum.mit.edu>
That mirrors what we're doing with autotools, and catches cases where
an earlier-tested flag affects the results for a later-tested flag
(e.g., -fwrapv can affect whether -fno-strict-overflow produces a
warning when tested).
Change-Id: I8ca0b6f5dee14d2394af9b6b8b93b9dbda2702d7
Reviewed-on: https://code.wireshark.org/review/3369
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Apply the patches printed out when the Debian package build complains,
and then remove the #MISSING indications that correspond to routines we
don't export outside the library (either because we don't declare them
as exported outside the library, because they've been removed, or
because they've been moved to another library).
Change-Id: Iba2d5c5436dabd31d7f84fd400bb78afcb5ee69f
Reviewed-on: https://code.wireshark.org/review/3367
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Add OS X content. Remove GTK+ content. Update the names in the NSIS
package to match the documentation (untested).
Change-Id: Id8fd08982bc26871bb8a319b0319808bcdba878c
Reviewed-on: https://code.wireshark.org/review/3366
Reviewed-by: Gerald Combs <gerald@wireshark.org>
For autotools, dump CC, CFLAGS, CXX, and CXXFLAGS.
For CMake, dump CMAKE_C_COMPILER_ID and CMAKE_CXX_COMPILER_ID.
Hopefully this will help figure out why CMake builds catch different
warnings from autotools builds.
Change-Id: I26955ad955f60e8bad248562fa87963a3a1bb42f
Reviewed-on: https://code.wireshark.org/review/3365
Reviewed-by: Guy Harris <guy@alum.mit.edu>
That way, we report the *full* contents of those flags.
Change-Id: Idab78049247c04b7d35a3ce804c8ea9daaf84312
Reviewed-on: https://code.wireshark.org/review/3364
Reviewed-by: Guy Harris <guy@alum.mit.edu>
I thought ws_symbol_export.h was supposed to eliminate API list files.
Apparently it didn't, so we have to update this file every time we add a
new API.
Change-Id: I93dda6ccc992c79d52d2b460e818995e6332f1be
Reviewed-on: https://code.wireshark.org/review/3363
Reviewed-by: Guy Harris <guy@alum.mit.edu>
This lets us blow up with oversized 64-bit length values, before casting
them to the 31-bit lengths we can actually handle in Wireshark, rather
than blindly casting them with weird results.
Use that in the MySQL dissector, and, if we get past the test, cast the
lengths to int to squelch warnings.
Change-Id: I3a5e9bd0027fa4ddcb9622f77952dba8f6b23c27
Reviewed-on: https://code.wireshark.org/review/3362
Reviewed-by: Guy Harris <guy@alum.mit.edu>
(this problem showed up in an APDU with two 16bit application ids)
Change-Id: Ie4842181b19db984a693534144fac5e91b217b34
Reviewed-on: https://code.wireshark.org/review/3358
Reviewed-by: Martin Kaiser <wireshark@kaiser.cx>
Tested-by: Martin Kaiser <wireshark@kaiser.cx>
I don't think it's an actual issue, but the new compiler on the mac buildbots
isn't smart enough to tell that.
Change-Id: I759e1afe9c4011d5612be0d15282076be6f5a331
Reviewed-on: https://code.wireshark.org/review/3355
Reviewed-by: Evan Huus <eapache@gmail.com>
In particular, epan/wslua/lrexlib.c has its own buffer_ routines,
causing some linker warnings on some platforms, as reported in bug
10332.
(Not to be backported to 1.12, as that would change the API and ABI of
libwsutil and libwiretap. We should also make the buffer_ routines in
epan/wslua/lrexlib.c static, which should also address this problem, but
the name change avoids other potential namespace collisions.)
Change-Id: I1d42c7d1778c7e4c019deb2608d476c52001ce28
Reviewed-on: https://code.wireshark.org/review/3351
Reviewed-by: Guy Harris <guy@alum.mit.edu>
What mystical new compiler upgrade is this?
Change-Id: I89b3bfb53b9a19bbfb1cc8339d38cdc4a4652c62
Reviewed-on: https://code.wireshark.org/review/3347
Reviewed-by: Evan Huus <eapache@gmail.com>
file in the build directory (rather than the source directory).
Change-Id: I365e573ee84e9a41aa76f4aa9a4a6efaf42fb60e
Reviewed-on: https://code.wireshark.org/review/3334
Reviewed-by: Evan Huus <eapache@gmail.com>
make-taps.pl needs to know where to find the source files otherwise none of
the tap data gets built correctly.
This makes the wslua test suite run in out-of-source-tree builds too.
Change-Id: I059474d90d59e87bd57dba18530a66a927a014cf
Reviewed-on: https://code.wireshark.org/review/3337
Reviewed-by: Evan Huus <eapache@gmail.com>
Apply a layout to the "Compiled Filter Output" dialog. Render the filter
as monospaced text.
Don't add an "expert OK" icon to the list. It doesn't seem to indicate
any sort of status.
Use a "Close" button instead of "OK". According to
http://uxmovement.com/buttons/why-the-ok-button-is-no-longer-okay/
"OK" implies we're going to perform some sort of action.
Add a "Copy" button while we're here.
Change-Id: Ia63f3032916e322fa9f4f2d465140b128b4f8078
Reviewed-on: https://code.wireshark.org/review/3338
Reviewed-by: Gerald Combs <gerald@wireshark.org>
Before, several managment packets were dissected incorrectly as
EPHandleDeleteReq's. Now they are dissected with the generic managment
packet dissector.
Change-Id: Id2f0951b91b99ba2340ff77c6285f382436788ef
Reviewed-on: https://code.wireshark.org/review/3328
Reviewed-by: Evan Huus <eapache@gmail.com>
If we have a pre-alternatives wireshark binary out there (e.g., because
we're upgrading from wireshark-1.10 to wireshark-1.12), get rid of it
in the %pre script. Otherwise the binary gets left there (because it's
listed as a %ghost in the new spec file).
Change-Id: I5828fcff486af3d269bad8dfe544c7659179bf5e
Reviewed-on: https://code.wireshark.org/review/3332
Reviewed-by: Jeff Morriss <jeff.morriss.ws@gmail.com>
Look for wpcap.lib in WpdPack/Lib/x64 when our target platform is Win64.
Change-Id: I9a1bac22106bcb6a1f155ce83c02e344e6ec0d55
Reviewed-on: https://code.wireshark.org/review/3331
Reviewed-by: Gerald Combs <gerald@wireshark.org>
Update some OS X information as well.
Change-Id: I9a252ec83712848696dbb531e6da15b3c7644591
Reviewed-on: https://code.wireshark.org/review/3327
Reviewed-by: Gerald Combs <gerald@wireshark.org>
It doesn't build on Linux but cmake looks for it (won't build without it) and
Windows users might want it.
Change-Id: I978f0de0a2895a82f4f3b8c1e9e0ecec6a93e6f4
Reviewed-on: https://code.wireshark.org/review/3325
Reviewed-by: Jeff Morriss <jeff.morriss.ws@gmail.com>
According to RFC 1323, the window scale shift value must not exceed 14.
Detect this and cap at 14 to prevent undefined behavior (shifting by a
too large value).
Caught by `clang -fsanitize=undefined`.
Change-Id: I1acad252b86c7f23e497575b48d9496346327e00
Reviewed-on: https://code.wireshark.org/review/3312
Reviewed-by: Michael Mann <mmann78@netscape.net>
Pascal Quantin