At least using MSYS2 python (that uses system() that uses CMD.EXE)
we must quote every command in a pipe, otherwise the "'C:' is not
recognized as an internal or external program" error occurs.
If we are using MSYS2 we use those packages to build Wireshark
using MinGW-w64 and disable most or all of our win-setup.ps1 codepaths.
Fix GLib configuration. Disable copying of DLLs with MSYS2.
Some tests in the suite_capture test suite are failing with MSYS
MINGW64. That particular set of tests is way too brittle regarding
file system paths; more work is needed to improve that situation.
This is more likely to hinder than to help, and is not really a good way to
invoke dpkg-buildpackage. It's a crutch that doesn't interact well with ninja,
has a 50/50 chance of giving the right results and will pollute the build
environment.
The Debian package should be built from a tarball instead.
The openSUSE 15.2 RPM Test step occasionally fails when trying to
retrieve repository metadata. Add "--no-remote" to the zypper install
command in the openSUSE 15.2 RPM Test to match the offline behavior of
our other RPM tests.
Implement little endian support for tvb_get_bits family of functions.
The big/little endian refers to bit numbering within an octet. In big
endian, the most significant bit is considered bit 0, while in little
endian the least significant bit is considered bit 0.
Add encoding parameters to proto tree bits format family functions.
Specify ENC_BIG_ENDIAN in all dissectors using these functions except in
USB HID that requires ENC_LITTLE_ENDIAN to work correctly.
When formatting bits values, always display most significant bit on the
leftmost position regardless of the encoding. This results in no gaps
between octets and makes the displayed value comprehensible.
Close#4478Fix#17014
The file type/subtype for built-in types are <=
wtap_num_builtin_file_types_subtypes - the plugin types are given
type/subtype values after the last built-in type/subtype value.
Fixes#17614.
We only allow exp-Golomb coded values to be as large as 32 bit
integers. When packets encode too large a value (invalid content),
clamp the value and report it as malformed with an expert info,
reporting the number of bits consumed (which will probably lead to
a BoundsError later in the packet.)
The case with 32 leading zeroes is a special case because for both
unsigned and signed interpretation there is one non overflowing value.
This is better than using DISSECTOR_ASSERT for invalid packet content.
Avoid left shifting a 32 bit integer by 32, which is undefined.
Use DISSECTOR_ASSERT_FIELD_TYPE at the beginning of the function rather
than using DISSECTOR_ASSERT in the middle, since it's more descriptive
in its error message and clearer code to do it at the start.
Same issue as #17612, commit a7dfe53488.
Special case the situation with 32 leading zeroes, since in C it's
undefined to left shift a 32 bit integer by 32. Only one value with
32 leading zeroes is an encoded 32 bit integer.
Clamp too large values to G_MAX[U]INT32 and report it as a malformed
expert info. Also report the supposed amount of bits consumed,
which will probably lead to a BoundsError down the line (possibly
not for some bit errors).
This is better than using DISSECTOR_ASSERT for invalid packet content.
Use DISSECTOR_ASSERT_FIELD_TYPE for doing the checks on the hf_field
types, since it's more descriptive in its error message.
Fix#17612.
Reverse the text added in cdd6f2ec80 and note that we can't yet use
Asciidoctor.js to build our documentation. I'm not sure how I managed to
miss this in my initial tests, but Asciidoctor.js is missing Docbook,
PDF, and EPUB backends, and doesn't support Ruby macros.
Provide Internet Archive links for dead URLs.
Update to note that PSML output is supported by tshark and not
a future feature (true since 17 years ago, when it was still tethereal).
Note "fake-field-wrapper" protocol for top level fields (including data,
which is converted from a protocol to a field for PDML).
Note "_ws.expert" protocol replaced by field, as with data.
Note that some dissectors place subdissected protocols in subtrees
instead of at the top level, and that this is _not_ changed, violating
the PDML spec.
Fix#10588.
Build output must not be placed in run/<config>/subdir.
This should be done using CMAKE_GENERATOR_IS_MULTI_CONFIG instead of just
MSVC but that wasn't working for me when I tried briefly.
This is applicable to every test (if we had more), not just HAVE_C99_VSNPRINTF.
Could also be a #define but let's go with this for now. This takes
advantage of the stack based design of CMakePushCheckState.
Capitalization matters here. We are now accepting this value from the
environment so we need to lower case that string.
Add an extra sanity check and status message for target CPU architecture
to help detect mismatches.
ETSI IP Datacast is another protocol that uses strings instead
of numeric payload types after the fmpt attribute. (ETSI TS 102 472,
ETSI TS 102 592). Since we're up to five special case strings to
check for, refactor the code a bit.
Protocol names can only be on the left hand side of filter expressions.
Commit ac0b1d42f3 (merge !4214) caused
unquoted strings that could be interpreted as either protocol names or
byte arrays to be parsed as the latter when on the RHS. Further
relax by allowing unquoted strings on the RHS that can be interpreted as
protocol names to be treated as any generic unquoted string. (The
semantic checker will still prefer interpreting the string as a byte
array, if possible, to a generic string.)
This is useful for filter expressions of the sort "frame contains data",
where data should be interpreted as "data", i.e. "\x64\x61\x74\x61".
Long run this ideally should be fixed earlier, at the lex parser or
grammar checker.
As per TS 12.21 section 8.1.4 "Manufacturer-Defined O&M messages",
NOTE 1:
"""
The Length Indicator gives the length of the Manufacturer-defined O&M data field in the message
segment being transported which is less than or equal to 255 octet.
"""
Where the "Manufacturer-defined O&M data field" is the content AFTER the
"ManId Length Indicator" + "Manuf. Identifier" as can be seen in the
table of the same section.
This fix was tested against osmo-bts, which implements the ipaccess
manufacturer extensions.
Profile files which is only used in Qt is not automatically registered
during startup and must be explicit registered.
Add profile_register_persconffile() to handle this registration.
Anders Broman
Pau Espin