* src/engine-gpgsm.c (status_handler):
* src/engine-uiserver.c (status_handler):
--
After a realloc (realloc is also used for initial alloc) the allocated
size if the buffer is not correctly recorded. Thus an overflow can be
introduced by receiving data with different line lengths in a specific
order. This is not easy exploitable because libassuan constructs the
line. However a crash has been reported and thus it might be possible
to constructs an exploit.
CVE-id: CVE-2014-3564
Reported-by: Tomáš Trnka
Resolved conflicts:
NEWS - removed
* src/w32-io.c (create_writer): Free CTX in cased of bad FD.
* src/w32-util.c (_gpgme_mkstemp): Free TMPNAME in case of a failed
mkstemp.
--
Found by Hans-Christoph Steiner with cppcheck.
* tests/gpg/start-stop-agent: New.
* tests/gpg/initial.test: New.
* tests/gpg/final.test: New.
* tests/gpg/Makefile.am (c_tests): New.
(TESTS): Move all to c_tests. Add initial.test, final.test, and
c_tests.
(TESTS_ENVIRONMENT): Add C_ALL=C
(private_keys): New.
(EXTRA_DIST): Add new files.
(./private-keys-v1.d/gpg-sample.stamp): Copy private keys.
(all-local): Depend on gpg-sample.stamp.
* tests/gpg/13CBE3758AFE42B5E5E2AE4CED27AFA455E3F87F: New.
* tests/gpg/13CD0F3BDF24BE53FE192D62F18737256FF6E4FD: New.
* tests/gpg/76F7E2B35832976B50A27A282D9B87E44577EB66: New.
* tests/gpg/7A030357C0F253A5BBCD282FFC4E521B37558F5C: New.
* tests/gpg/A0747D5F9425E6664F4FFBEED20FBCA79FDED2BD: New.
--
GnuPG 2.1 requires that the agent knows the private keys. Thus we
need to install them. The code should still work with GnuPG < 2.1.
To make the whole testing using a agent more robust, the way to start
the agent has also be changed.
* configure.ac: Move test for gpgconf before test for gpg.
(GPG, GPGSM, G13): Use gpgconf instead of AC_PATH_PROG if possible.
--
This change is required so that the t-engine-info test does not
fail. In any case, it is a good idea to use gpgconf for these test
because this is how it works at runtime.
* src/gpgme.h.in (gpgme_data_type_t): New.
(gpgme_data_identify): New prototype.
* src/data-identify.c: New.
* src/parsetlv.c, src/parsetlv.h: New. Take from gpa.
* src/libgpgme.vers, src/gpgme.def: Add gpgme_data_identify.
* src/gpgme-tool.c (status): Add STATUS_IDENTIFY_RESULT.
(gt_identify): New.
(cmd_identify): New.
(hlp_passwd): Move close to cmd_passwd.
--
It is often useful to have a way to identify the data which needs
processing. This is such a common task that it makes sense to
implement this in gpgme to avoid diverging implementations.
* src/util.h: Move some prototypes to ...
* src/sys-util.h: New.
* src/Makefile.am (main_sources): Add sys-util.h.
* configure.ac (AH_VERBATIM): Add DIRSEP_C and DIRSEP_S.
* src/dirinfo.c: Include sys-util.h.
(WANT_GPG_NAME, WANT_GPGSM_NAME, WANT_G13_NAME)
(WANT_UISRV_SOCKET): New.
(dirinfo): Add corresponding fields.
(parse_output): Add arg COMPONENTS and set new fields.
(read_gpgconf_dirs): Add arg components and act upon it.
(get_gpgconf_item): Call read_gpgconf_dirs two times. Add debug
output.
(_gpgme_get_default_gpg_name): New.
(_gpgme_get_default_gpgsm_name): New.
(_gpgme_get_default_g13_name): New.
(_gpgme_get_default_gpgconf_name): New.
(_gpgme_get_default_uisrv_socket): New.
* src/engine-gpg.c, src/engine-g13.c, src/engine-gpgconf.c
* src/engine-gpgsm.c, src/engine-uiserver.c: Change to use
_gpgme_get_default_ instead of those from sys-util.h.
* src/posix-util.c (_gpgme_get_gpg_path): Include sys-util.h.
(_gpgme_get_uiserver_socket_path): Remove.
* src/w32-util.c (_gpgme_get_gpg_path): Include sys-util.h.
(_gpgme_get_uiserver_socket_path): Remove.
--
The default engines names are now taken from the output of gpgconf.
If gpgconf is not installed gpg 1 is assumed and locate the same was
as gpgconf.
* build-aux/ltmain.sh (sed_uncomment_deffile): New.
(orig_export_symbols): Uncomment def file before testing for EXPORTS.
* m4/libtool.m4: Do the same for the generated code.
--
The old code was not correct in that it only looked at the first line
and puts an EXPORTS keyword in front if missing. Binutils 2.22
accepted a duplicated EXPORTS keyword but at least 2.23.2 is more
stringent and bails out without this fix.
There is no need to send this upstream. Upstream's git master has a
lot of changes including a similar fix for this problems. There are
no signs that a libtool 2.4.3 will be released to fix this problem and
thus we need to stick to our copy of 2.4.2 along with this patch.
Signed-off-by: Werner Koch <wk@gnupg.org>
* src/gpgme.h.in: Move gpgme_sssize_t and gpgme_off_t typedefs into
the extern "C" scope.
--
Frankly, this does not help very much because g++ still does not
interpret gpgme_data_seek_cb_t as compatible with the C++ ssize_t. I
am sorry for that API change but C and C++ are too different to always
get both of them under one umbrella. However, the ABI should now be
identical to older gpgme versions.
* configure.ac (NEED__FILE_OFFSET_BITS): Change to define gpgme_off_t
and gpgme_ssize_t.
(API__OFF_T, API__SSIZE_T): New ac_subst.
* src/gpgme.h.in: Replace all ssize_t and off_t by ac_subst macros.
* src/assuan-support.c, src/ath-pthread.c, src/ath.c, src/ath.h
* src/data-compat.c, src/data-fd.c, src/data-mem.c, src/data-stream.c
* src/data-user.c, src/data.c, src/data.h, src/engine-gpgsm.c
* src/engine-uiserver.c, src/gpgme-tool.c, src/gpgme.c: Replace off_t
by gpgme_off_t and sszie_t by gpgme_ssize_t.
* src/ath-pthread.c, src/ath.h: Include gpgme.h.
--
For a detailed description, see the gpgme.texi diff.
* src/w32-io.c (reader_context_s, create_reader)
(writer_context_s, create_writer): Rename STOPPED to CLOSE_EV.
(reader, writer): Remove setting of STOPPED. Wait for CLOSE_EV and
then release the context.
(destroy_reader, destroy_writer): Do not wait but set the CLOSE_EV.
(kill_reader, kill_writer): Remove.
(_gpgme_io_close): Add code from kill_reader and kill_writer.
--
The old code was prone to deadlocks which were actually exhibited at
Kleopatra startup. The new code is much more straightforward and
easier to understand. The reason for the complex old code was
probably due to our former idea to allow re-use of the I/O threads.
However we have long given up on this.
* configure.ac: Disable fd-passing by default for Apple.
--
We have not yet tracked down the problem, thus we revert to the
pre-1.4 behaviour for Apple.
GnuPG-bug-id: 1483
* configure.ac (INSERT__TYPEDEFS_FOR_GPGME_H): Add hacks for 32 bit
mingw64.
* src/util.h [W32]: Include winsock2.h before windows to make mingw64
happy.
* src/w32-util.c (_WIN32_IE): Need to use 5.1 for mingw64.
--
It is not clear to me what mingw64 wants to achieve with their POSIX
hacks. In particular the off64_t stuff looks quite strange given that
Windows has a pretty stable API for close to 2 decades. Thus I can't
say whether it will really work when build with that toolchain.
* configure.ac: Rename my_foo variables to mym4_foo variables to make
clear that they are processed by m4.
(VERSION_NUMBER): New ac_subst.
(AH_BOTTOM): Add CRIGHTBLURB macro.
(BUILD_REVISION, BUILD_FILEVERSION, BUILD_TIMESTAMP): Change them to
work similar to libgpg-error et al.
* src/versioninfo.rc.in: Remove use of BUILD_NUMBER and get it in line
with gpg-error et al.
* src/version.c (cright_blurb): New.
(gpgme_check_version_internal): Add magic to display the above
information.
* tests/t-version.c (main): Add option --verbose.
* configure.ac: Use AC_CONFIG_AUX_DIR. Remove args from
AM_INIT_AUTOMAKE. Replace AM_CONFIG_HEADER by AC_CONFIG_HEADER.
* compile, config.guess, config.sub, depcomp, install-sh, ltmain.sh
* mkinstalldirs, texinfo.texi: Move to build-aux/ and update from
gnulib (c042abf).
* build-aux/mdate-sh, build-aux/missing: Install via automake -a -c.
Update autogen.sh for changed config dir.
* autogen.sh: Adjust for scripts dir change. Update W32 toolprefix
list.
