platform/upstream/dbus.git
13 years agoPrepare version 1.4.4 dbus-1.4.4
Simon McVittie [Thu, 17 Feb 2011 18:21:25 +0000 (18:21 +0000)]
Prepare version 1.4.4

13 years agoSplit acinclude.m4 into separate files in m4/
Simon McVittie [Thu, 17 Feb 2011 18:21:16 +0000 (18:21 +0000)]
Split acinclude.m4 into separate files in m4/

Drop AM_CHECK_PYTHON_HEADERS (no longer used) and split the others
according to their upstreams.

Otherwise, compiler.m4 won't get included in tarballs.

13 years agoMerge branch 'leak-fixes-33126'
Simon McVittie [Thu, 17 Feb 2011 17:19:13 +0000 (17:19 +0000)]
Merge branch 'leak-fixes-33126'

Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126

13 years agoImport compiler.m4 and lcov.am from telepathy-glib, and use them to replace gcov
Simon McVittie [Wed, 2 Feb 2011 16:51:30 +0000 (16:51 +0000)]
Import compiler.m4 and lcov.am from telepathy-glib, and use them to replace gcov

Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=10887

13 years agoDon't ignore all files in m4, just the libtool ones
Simon McVittie [Mon, 10 Jan 2011 15:58:42 +0000 (15:58 +0000)]
Don't ignore all files in m4, just the libtool ones

Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=10887

13 years agoDon't distribute dbus-daemon.1 in the tarball
Simon McVittie [Mon, 7 Feb 2011 12:00:31 +0000 (12:00 +0000)]
Don't distribute dbus-daemon.1 in the tarball

It's generated by configure.

Also use a more normal way to distribute man pages that are actually
source, and use man1_MANS rather than auto-detecting the section from
the man page, which is best-practice and slightly faster.

Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=34292

13 years agoDon't distribute dbus-arch-deps.h in the tarball
Simon McVittie [Mon, 7 Feb 2011 11:58:35 +0000 (11:58 +0000)]
Don't distribute dbus-arch-deps.h in the tarball

It's generated by configure and specific to a system, so there's no point.

Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=34292

13 years agoRemove references to dbus-md5, which hasn't been used for at least 8 years
Simon McVittie [Mon, 7 Feb 2011 11:57:57 +0000 (11:57 +0000)]
Remove references to dbus-md5, which hasn't been used for at least 8 years

Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=34292

13 years agoDBusTransport: don't copy DBusAuth's GUID to expected_guid
Simon McVittie [Tue, 1 Feb 2011 12:27:39 +0000 (12:27 +0000)]
DBusTransport: don't copy DBusAuth's GUID to expected_guid

There was previously a leak in _dbus_transport_get_is_authenticated, if
we hit OOM when copying the server's actual GUID. This would fail the
"SHA1 connection test" if _dbus_iterate was modified to allocate and free
one extra pointer per iteration.

Rather than adding OOM handling everywhere, if we avoid strdup'ing the
GUID, and just re-fetch the const string from the DBusAuth object on demand
instead, we go back to a situation where _dbus_transport_get_is_authenticated
can't fail.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126

13 years agoupdate_desktop_file_entry: stylistic fixes based on Colin's review
Simon McVittie [Tue, 1 Feb 2011 13:35:33 +0000 (13:35 +0000)]
update_desktop_file_entry: stylistic fixes based on Colin's review

13 years agoupdate_desktop_file_entry: unify cleanup code for success and failure cases
Simon McVittie [Mon, 17 Jan 2011 11:46:09 +0000 (11:46 +0000)]
update_desktop_file_entry: unify cleanup code for success and failure cases

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126

13 years agoupdate_desktop_file_entry: don't double-free strings if added to entry before failure
Simon McVittie [Mon, 17 Jan 2011 11:45:47 +0000 (11:45 +0000)]
update_desktop_file_entry: don't double-free strings if added to entry before failure

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126

13 years agoupdate_desktop_file_entry: free @exec on error
Simon McVittie [Mon, 17 Jan 2011 11:42:48 +0000 (11:42 +0000)]
update_desktop_file_entry: free @exec on error

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126

13 years agoupdate_desktop_file_entry: make scope of exec_tmp as short as possible
Simon McVittie [Fri, 14 Jan 2011 17:59:24 +0000 (17:59 +0000)]
update_desktop_file_entry: make scope of exec_tmp as short as possible

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126

13 years agoupdate_desktop_file_entry: use _dbus_strdup for something we'll dbus_free
Simon McVittie [Fri, 14 Jan 2011 17:59:10 +0000 (17:59 +0000)]
update_desktop_file_entry: use _dbus_strdup for something we'll dbus_free

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126

13 years agoNEWS so far for 1.4.4
Simon McVittie [Wed, 16 Feb 2011 12:37:59 +0000 (12:37 +0000)]
NEWS so far for 1.4.4

13 years agobuild-sys: ignore -Waddress message so that build works on gcc/rawhide
Lennart Poettering [Tue, 15 Feb 2011 12:29:28 +0000 (13:29 +0100)]
build-sys: ignore -Waddress message so that build works on gcc/rawhide

Newer gccs warn if you compare an address of a variable that is allocated
on the stack or is static with NULL. Since we compile dbus with -Werror
this causes the build to fail since we do this check all the time due to
macros such as _DBUS_ASSERT_ERROR_IS_SET().

13 years agoFix syslog string processing
Matthias Clasen [Mon, 14 Feb 2011 16:30:11 +0000 (11:30 -0500)]
Fix syslog string processing

squash_string_nonprintable didn't actually modify the string.

Reviewed-By: Colin Walters <walters@verbum.org>
13 years agobus: Raise file descriptor limit to match configuration
Colin Walters [Mon, 31 Jan 2011 20:22:14 +0000 (15:22 -0500)]
bus: Raise file descriptor limit to match configuration

The default configuration has hardcoded 2048 complete connections,
and 64 incomplete.  We need at least that number of file descriptors,
plus some for internal use.

In the bus, attempt to call setrlimit() before we drop privileges.
Practically speaking for this means the system bus gets it, the
session bus doesn't.

http://bugs.freedesktop.org/show_bug.cgi?id=33474

Reviewed-By: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoMerge branch 'dist-30285' into dbus-1.4
Simon McVittie [Wed, 2 Feb 2011 15:15:01 +0000 (15:15 +0000)]
Merge branch 'dist-30285' into dbus-1.4

Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agoMerge branch 'serverdata-33277' into dbus-1.4
Simon McVittie [Wed, 2 Feb 2011 15:14:29 +0000 (15:14 +0000)]
Merge branch 'serverdata-33277' into dbus-1.4

Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agoIgnore files produced by `make dist`
Simon McVittie [Tue, 1 Feb 2011 18:04:35 +0000 (18:04 +0000)]
Ignore files produced by `make dist`

13 years agoDon't distribute generated files that embed our builddir
Simon McVittie [Tue, 1 Feb 2011 18:04:21 +0000 (18:04 +0000)]
Don't distribute generated files that embed our builddir

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=30285

13 years agobus-test: add support for only running one test
Simon McVittie [Fri, 14 Jan 2011 15:16:12 +0000 (15:16 +0000)]
bus-test: add support for only running one test

This is much quicker when valgrinding.

Reviewed-by: Colin Walters <walters@verbum.org>
13 years agodbus_bus_set_unique_name, dbus_bus_get_unique_name: remember to unlock on OOM
Simon McVittie [Tue, 1 Feb 2011 11:41:02 +0000 (11:41 +0000)]
dbus_bus_set_unique_name, dbus_bus_get_unique_name: remember to unlock on OOM

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33126
Reviewed-by: Colin Walters <walters@verbum.org>
13 years agoIf a file included via <includedir/> is invalid, syslog and skip it
Simon McVittie [Mon, 31 Jan 2011 18:06:09 +0000 (18:06 +0000)]
If a file included via <includedir/> is invalid, syslog and skip it

This is more graceful than failing entirely (our previous behaviour),
but more visible than ignoring it completely (the previous behaviour
patched in by Debian and derivatives).

Based on a patch from Daniel Silverstone back in 2004, which was meant
to be temporary; I think it makes sense to change this permanently,
since files in *.d are typically supplied by other packages, whose bugs
shouldn't be able to bring down dbus-daemon.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=19186
Bug-Debian: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=230231
Reviewed-by: Colin Walters <walters@verbum.org>
13 years agoImplement message type in match_rule_to_string
Alban Crequy [Wed, 6 Oct 2010 11:12:32 +0000 (13:12 +0200)]
Implement message type in match_rule_to_string

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=24307
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoOpt-in to thread safety on Solaris
Simon McVittie [Wed, 26 Jan 2011 18:21:49 +0000 (18:21 +0000)]
Opt-in to thread safety on Solaris

Alternatively, get a better C library. :-)

See also:
<http://lists.freedesktop.org/archives/dbus/2010-November/013796.html>,
<http://lists.freedesktop.org/archives/dbus/2010-December/013902.html>.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33464
Reviewed-by: Colin Walters <walters@verbum.org>
13 years ago_dbus_server_new_for_tcp_socket: fix error handling
Simon McVittie [Wed, 26 Jan 2011 18:55:00 +0000 (18:55 +0000)]
_dbus_server_new_for_tcp_socket: fix error handling

This is one of the patches from Maemo's dbus package. It seems to do all
of:

* fix some documentation
* remove unreached code to delete/free the nonce file from
  _dbus_server_new_for_socket - doing that on failure violates
  least-astonishment anyway
* in _dbus_server_new_for_tcp_socket, never fail without setting @error
* if we fail after creating the nonce file, delete it
* if we fail after allocating the nonce file struct, free it

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoMake _dbus_directory_get_next_file use readdir, not readdir_r
Simon McVittie [Mon, 17 Jan 2011 18:40:15 +0000 (18:40 +0000)]
Make _dbus_directory_get_next_file use readdir, not readdir_r

This isn't thread-safe or reentrant, but it turns out we don't need
either of those properties, and readdir_r is a real pain to use correctly,
particularly in the presence of FUSE filesystems that might implement
statfs() wrong.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=8284
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=15922
Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agotest/name-test: remove dangling references to test-names.c, which is absent
Simon McVittie [Tue, 18 Jan 2011 18:07:14 +0000 (18:07 +0000)]
test/name-test: remove dangling references to test-names.c, which is absent

Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agofix check for 32-bit integers if short is 32 bits long
Simon McVittie [Wed, 19 Jan 2011 14:09:33 +0000 (14:09 +0000)]
fix check for 32-bit integers if short is 32 bits long

In practice, this won't happen for years, but we might as well be correct.

Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
13 years ago_dbus_loop_iterate: if the kernel says a fd is bad, stop watching it
Simon McVittie [Wed, 12 Jan 2011 12:14:49 +0000 (12:14 +0000)]
_dbus_loop_iterate: if the kernel says a fd is bad, stop watching it

Again, this shouldn't happen - modules are responsible for cleaning up
their watches - but the failure mode here is really bad: if we leave an
invalid fd in the set, every poll() call will instantly return, marking
it as POLLNVAL. The result is that dbus-daemon busy-loops on poll()
without responding to I/O, so the bad watch will probably never be
cleared up.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32992
Bug-NB: NB#200248
Reviewed-by: Colin Walters <walters@verbum.org>
13 years ago_dbus_loop_iterate: cleanup: make more use of a temporary variable
Simon McVittie [Wed, 12 Jan 2011 11:59:19 +0000 (11:59 +0000)]
_dbus_loop_iterate: cleanup: make more use of a temporary variable

It was added by the previous commit.

Reviewed-by: Colin Walters <walters@verbum.org>
13 years ago_dbus_loop_iterate: if a watch has been invalidated, skip it
Simon McVittie [Wed, 12 Jan 2011 11:58:31 +0000 (11:58 +0000)]
_dbus_loop_iterate: if a watch has been invalidated, skip it

This shouldn't happen - other modules are responsible for cleaning up
their watches - but the bug fixed in my last commit has been present for
several years and I'm sure it's not the only one, so for robustness,
let's refuse to watch obviously-wrong file descriptors.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32992
Bug-NB: NB#200248
Reviewed-by: Colin Walters <walters@verbum.org>
13 years agodbus-spawn: don't leave bad file descriptors being watched
Simon McVittie [Tue, 11 Jan 2011 19:57:31 +0000 (19:57 +0000)]
dbus-spawn: don't leave bad file descriptors being watched

The code called from handle_watch() might close either or both of the
sockets we're watching, without cleaning up the DBusWatch. This results
in invalid file descriptors being passed to _dbus_poll(), which could
end up busy-looping on a POLLNVAL condition until the babysitter loses
its last ref (which automatically clears up both watches).

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32992
Bug-NB: NB#200248
Reviewed-by: Colin Walters <walters@verbum.org>
13 years agotest-privserver-client: wait for server to die between iterations
Simon McVittie [Thu, 20 Jan 2011 13:41:58 +0000 (13:41 +0000)]
test-privserver-client: wait for server to die between iterations

This fixes a race condition: the server exits while the client continues
to the next iteration. If the server wins, the test passes. If the client
wins, it sends a message to the dying service, never gets a reply, and the
test fails.

My branch to refactor the main loop for fd.o #23194 seems to make the
client more likely to win this race, resulting in intermittent test
failures.

This is an instance of the general problem described by fd.o #11454.

13 years agotest_server_shutdown: disconnect the server before causing it to be freed
Simon McVittie [Wed, 19 Jan 2011 17:34:10 +0000 (17:34 +0000)]
test_server_shutdown: disconnect the server before causing it to be freed

Otherwise we die with an assertion failure.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33277

13 years agotest_server_setup: allocate two server-data blobs, for watches and timeouts
Simon McVittie [Wed, 19 Jan 2011 17:33:31 +0000 (17:33 +0000)]
test_server_setup: allocate two server-data blobs, for watches and timeouts

This is similar to how ConnectionData works. Without this change, we
deserve to segfault: when the first set of callbacks (either watches or
timeouts) is cleaned up, we unref the server and loop, and free sd;
when the second set of callbacks is cleaned up, we use-after-free sd,
the server and the loop, then double-free sd.

However, due to fd.o #33277 we don't even get that far, because we've
already died with an assertion failure.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33277

13 years agotest_server_setup: on OOM while setting up watch/timeout, don't leak sd
Simon McVittie [Wed, 19 Jan 2011 17:28:58 +0000 (17:28 +0000)]
test_server_setup: on OOM while setting up watch/timeout, don't leak sd

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Reviwed-by: Colin Walters <walters@verbum.org>
13 years ago_dbus_read_credentials_socket: support OpenBSD
Simon McVittie [Fri, 21 Jan 2011 13:27:18 +0000 (13:27 +0000)]
_dbus_read_credentials_socket: support OpenBSD

OpenBSD SO_PEERCRED is nearly the same as Linux SO_PEERCRED, but the
struct is named differently. Sigh.

Based on a patch from 'ajacoutot'.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32542

13 years agofix bad DBusMessageIter doc example
Richard A. Hankins [Wed, 19 Jan 2011 15:59:24 +0000 (15:59 +0000)]
fix bad DBusMessageIter doc example

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=11190
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agosystemd: ensure that the system bus is unlikely to be killed on OOM
Lennart Poettering [Tue, 18 Jan 2011 22:46:15 +0000 (23:46 +0100)]
systemd: ensure that the system bus is unlikely to be killed on OOM

13 years ago_dbus_listen_tcp_socket: unify cleanup for EADDRINUSE and other failures
Simon McVittie [Tue, 18 Jan 2011 16:07:24 +0000 (16:07 +0000)]
_dbus_listen_tcp_socket: unify cleanup for EADDRINUSE and other failures

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=29881

13 years ago_dbus_listen_tcp_socket: free listen_fd in the error case
Christian Dywan [Tue, 18 Jan 2011 16:00:05 +0000 (16:00 +0000)]
_dbus_listen_tcp_socket: free listen_fd in the error case

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=29881
Bug-NB: NB#180486
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoFix failure in configure test for abstract sockets
Wulf C. Krueger [Mon, 30 Aug 2010 22:45:12 +0000 (22:45 +0000)]
Fix failure in configure test for abstract sockets

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=29895
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoclean up credentials on OOM in process_config_first_time_only
Christian Dywan [Mon, 17 Jan 2011 12:41:12 +0000 (12:41 +0000)]
clean up credentials on OOM in process_config_first_time_only

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Bug-NB: NB#180486
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agodbus-cleanup-sockets: free list of socket entries
Simon McVittie [Mon, 17 Jan 2011 12:18:40 +0000 (12:18 +0000)]
dbus-cleanup-sockets: free list of socket entries

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Bug-NB: NB#180486
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoremove obviously-redundant code from list_concat_new
Simon McVittie [Mon, 17 Jan 2011 12:04:09 +0000 (12:04 +0000)]
remove obviously-redundant code from list_concat_new

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Bug-NB: NB#180486
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agohandle failure to register inotify shutdown function
Christian Dywan [Mon, 17 Jan 2011 12:02:22 +0000 (12:02 +0000)]
handle failure to register inotify shutdown function

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Bug-NB: NB#180486
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agobus_matchmaker_new: free matchmaker itself, on OOM
Christian Dywan [Mon, 17 Jan 2011 11:58:23 +0000 (11:58 +0000)]
bus_matchmaker_new: free matchmaker itself, on OOM

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Bug-NB: NB#180486
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agodbus-nonce: check results of _dbus_string_init for OOM
Christian Dywan [Mon, 17 Jan 2011 11:53:57 +0000 (11:53 +0000)]
dbus-nonce: check results of _dbus_string_init for OOM

Origin: vendor, Maemo
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=33128
Bug-NB: NB#180486
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoMerge branch 'tmp'
Simon McVittie [Fri, 14 Jan 2011 15:19:18 +0000 (15:19 +0000)]
Merge branch 'tmp'

Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32854

13 years agoRemove vestigial support for $INTLLIBS
Simon McVittie [Thu, 6 Jan 2011 18:21:31 +0000 (18:21 +0000)]
Remove vestigial support for $INTLLIBS

We no longer call AM_GLIB_GNU_GETTEXT or use gettext directly.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=25306
Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agoAdd autolaunch test to .gitignore
Simon McVittie [Thu, 6 Jan 2011 17:35:35 +0000 (17:35 +0000)]
Add autolaunch test to .gitignore

13 years agocmake/cross-compile.sh: use mktemp(1) to avoid a symlink attack in /tmp
Simon McVittie [Wed, 5 Jan 2011 19:10:43 +0000 (19:10 +0000)]
cmake/cross-compile.sh: use mktemp(1) to avoid a symlink attack in /tmp

13 years agoFix the check for -Wfloat-equal
Simon McVittie [Wed, 5 Jan 2011 19:10:24 +0000 (19:10 +0000)]
Fix the check for -Wfloat-equal

13 years agoconfigure.in: use AC_TRY_COMPILE to avoid a symlink attack in /tmp during compilation
Simon McVittie [Wed, 5 Jan 2011 19:10:12 +0000 (19:10 +0000)]
configure.in: use AC_TRY_COMPILE to avoid a symlink attack in /tmp during compilation

13 years agoFree session file early in dbus-launch
Christian Dywan [Thu, 23 Sep 2010 16:21:25 +0000 (09:21 -0700)]
Free session file early in dbus-launch

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=29881
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoVerify that getsockname succeeded in _dbus_listen_tcp_socket
Christian Dywan [Mon, 19 Jul 2010 11:05:29 +0000 (13:05 +0200)]
Verify that getsockname succeeded in _dbus_listen_tcp_socket

Bug: https://bugs.freedesktop.org//show_bug.cgi?id=29881
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoHandle failure to allocate error message in _read_subprocess_line_argv
Christian Dywan [Mon, 19 Jul 2010 09:36:15 +0000 (11:36 +0200)]
Handle failure to allocate error message in _read_subprocess_line_argv

Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=29881

13 years agoCheck return value of XGetWindowProperty in x11_get_address
Christian Dywan [Mon, 19 Jul 2010 09:48:06 +0000 (11:48 +0200)]
Check return value of XGetWindowProperty in x11_get_address

Reviewed-by: Thiago Macieira <thiago@kde.org>
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=29881

13 years agoAnnotate dbus_malloc*() functions with __attribute__ malloc and __attribute__ alloc_size
Cristian Rodríguez [Wed, 29 Dec 2010 00:10:31 +0000 (21:10 -0300)]
Annotate dbus_malloc*() functions with __attribute__ malloc and __attribute__ alloc_size

This improves optimization as well catches buffer overflows at compile time.

code like:

foo = dbus_malloc(2);
strcpy(foo, "epicfail");

will emit:

"warning ... will always overflow the destination buffer..."

Signed-off-by: Cristian Rodríguez <cristian.rodriguez@opensuse.org>
Reviewed-by: Thiago Macieira <thiago@kde.org>
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32710

13 years agoconnection: correct HAVE_DECL_MSG_NOSIGNAL guard
Will Thompson [Fri, 24 Dec 2010 14:40:49 +0000 (14:40 +0000)]
connection: correct HAVE_DECL_MSG_NOSIGNAL guard

Helpfully, AC_CHECK_DECLS (which configure.in uses to define
HAVE_DECL_MSG_NOSIGNAL) behaves differently to every similar AC_CHECK_*
macro and, rather than producing #undef HAVE_DECL_MSG_NOSIGNAL if the
given symbol is not found, instead produces #define
HAVE_DECL_MSG_NOSIGNAL 0.

Mike McQuaid's patch fixes the uses of this constant in sysdeps-unix;
with this patch, all code is guarded consistently and correctly.

13 years agoFix error condition in _dbus_pipe_close
Christian Dywan [Tue, 4 Jan 2011 19:36:43 +0000 (19:36 +0000)]
Fix error condition in _dbus_pipe_close

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32264
Reviewed-by: Colin Walters <walters@verbum.org>
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoMake _dbus_system_logv actually exit with DBUS_SYSTEM_LOG_FATAL
Christian Dywan [Tue, 4 Jan 2011 19:23:47 +0000 (19:23 +0000)]
Make _dbus_system_logv actually exit with DBUS_SYSTEM_LOG_FATAL

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32262
Bug-NB: NB#180486
Reviewed-by: Thiago Macieira <thiago@kde.org>
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
13 years agoWindows fix: MSVC 2010 has errno.h.
Pau Garcia i Quiles [Fri, 31 Dec 2010 15:10:42 +0000 (16:10 +0100)]
Windows fix: MSVC 2010 has errno.h.

13 years agoAvoid live lock in Windows (CE) under some situations due to unfair condition variables.
Marcus Brinkmann [Wed, 22 Dec 2010 16:02:33 +0000 (17:02 +0100)]
Avoid live lock in Windows (CE) under some situations due to unfair condition variables.

13 years agoUse absolute session service dir path in bus-test on windows.
Ralf Habacker [Wed, 29 Dec 2010 11:31:42 +0000 (12:31 +0100)]
Use absolute session service dir path in bus-test on windows.

13 years agoKeep in sync with autotools.
Ralf Habacker [Wed, 29 Dec 2010 13:41:41 +0000 (14:41 +0100)]
Keep in sync with autotools.

13 years agosysdeps-unix: use #if for HAVE_DECL_MSG_NOSIGNAL
Mike McQuaid [Fri, 24 Dec 2010 13:30:22 +0000 (13:30 +0000)]
sysdeps-unix: use #if for HAVE_DECL_MSG_NOSIGNAL

Signed-off-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agoMention maintainer-upload-docs in the release process
Will Thompson [Tue, 21 Dec 2010 15:56:34 +0000 (15:56 +0000)]
Mention maintainer-upload-docs in the release process

13 years agoBump development version to 1.4.3.
Will Thompson [Tue, 21 Dec 2010 15:45:03 +0000 (15:45 +0000)]
Bump development version to 1.4.3.

The last release should have been 1.4.2; sorry.

13 years agoFix configure doc: libxml may be named libxml2 on some systems.
Ralf Habacker [Tue, 21 Dec 2010 07:46:52 +0000 (08:46 +0100)]
Fix configure doc: libxml may be named libxml2 on some systems.

13 years agoKeep in sync with automake.
Ralf Habacker [Tue, 21 Dec 2010 07:24:00 +0000 (08:24 +0100)]
Keep in sync with automake.

13 years agoDist additional README's too.
Ralf Habacker [Tue, 21 Dec 2010 07:23:27 +0000 (08:23 +0100)]
Dist additional README's too.

13 years agoPost-release version bump to 1.4.2
Will Thompson [Mon, 20 Dec 2010 21:24:24 +0000 (21:24 +0000)]
Post-release version bump to 1.4.2

13 years agoBump LT_REVISION for this new release. dbus-1.4.1
Will Thompson [Mon, 20 Dec 2010 21:18:17 +0000 (21:18 +0000)]
Bump LT_REVISION for this new release.

13 years agoNEWS for 1.4.1
Will Thompson [Mon, 20 Dec 2010 21:18:10 +0000 (21:18 +0000)]
NEWS for 1.4.1

13 years agoCVE 2010-4352: Reject deeply nested variants
Havoc Pennington [Mon, 13 Dec 2010 02:08:43 +0000 (21:08 -0500)]
CVE 2010-4352: Reject deeply nested variants

Add DBUS_INVALID_NESTED_TOO_DEEPLY validity problem and a test that
should generate it.

Previously, we rejected deep nesting in the signature, but
variants allow dynamic message nesting, conditional only
on the depth of the message body.

The nesting limit is 64, which was also the limit in static
signatures.  Empirically, dynamic nesting depth observed on my
Fedora 14 system doesn't exceed 2; 64 is really a huge limit.

https://bugs.freedesktop.org/show_bug.cgi?id=32321

Signed-Off-By: Colin Walters <walters@verbum.org>
Signed-off-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agopolicy tests: Use bin rather than nogroup, wheel, or root
Will Thompson [Fri, 17 Dec 2010 18:42:13 +0000 (18:42 +0000)]
policy tests: Use bin rather than nogroup, wheel, or root

This group seems to exist on all of Debian, Fedora and Mac OS X. (I
mistakenly believed that nogroup exists on Fedora.)

13 years agosysdeps-unix: check for HAVE_DECL_MSG_NOSIGNAL
Will Thompson [Thu, 16 Dec 2010 13:56:04 +0000 (13:56 +0000)]
sysdeps-unix: check for HAVE_DECL_MSG_NOSIGNAL

MSG_NOSIGNAL could theoretically be an enum member or something rather
than an #define, so it's better to check for the flag defined by the
configure check (as is done in dbus-connection.c already).

Reviewed-by: Colin Walters <walters@verbum.org>
Reviewed-by: Scott James Remnant <scott@netsplit.com>
13 years agosysdeps-unix: use MSG_NOSIGNAL when sending creds
Scott James Remnant [Thu, 16 Dec 2010 11:11:44 +0000 (11:11 +0000)]
sysdeps-unix: use MSG_NOSIGNAL when sending creds

Since SIGPIPE is no longer touched by default when MSG_NOSIGNAL is
available, it's extra-critical that all socket writes actually pass
that flag.

Signed-off-by: Will Thompson <will.thompson@collabora.co.uk>
13 years agoRevert "test-autolaunch is only known to work in launchd environments."
Will Thompson [Mon, 13 Dec 2010 17:54:57 +0000 (17:54 +0000)]
Revert "test-autolaunch is only known to work in launchd environments."

This reverts commit 4626b405609900812125a407ef5fdd31c7b89868.
test-autolaunch works fine in non-launchd environments (and non-X11
environments, based on a quick test passing enable_x11=no to configure).
On the contrary: this commit *broke* the build on non-launchd
environments, because test/name-test/run-test.sh still tried to run this
test even if it hadn't been built.

13 years agoIgnore generated bus/org.freedesktop.dbus-session.plist
Will Thompson [Mon, 13 Dec 2010 17:52:40 +0000 (17:52 +0000)]
Ignore generated bus/org.freedesktop.dbus-session.plist

13 years agopolicy tests: Use 'nogroup' rather than 'wheel' or 'root'
Will Thompson [Mon, 13 Dec 2010 17:22:00 +0000 (17:22 +0000)]
policy tests: Use 'nogroup' rather than 'wheel' or 'root'

The excellently-titled commit 197bef8 “Fix test failures on OSX.” broke
the tests on Linux, since there's no wheel group on this side of the
tracks. So here's a group everyone should enjoy.

(If anyone comes along and tells me that DragonflyBSD doesn't have
'nogroup' …)

13 years agoHandle case when autolaunched daemon address is already pulished on windows.
Ralf Habacker [Tue, 14 Dec 2010 07:12:59 +0000 (08:12 +0100)]
Handle case when autolaunched daemon address is already pulished on windows.

13 years agoWindows fix: In install path scope case add hashed install path to mutex/shm name.
Ralf Habacker [Mon, 13 Dec 2010 21:17:12 +0000 (22:17 +0100)]
Windows fix: In install path scope case add hashed install path to mutex/shm name.

13 years agoFix malformed specification XML.
Will Thompson [Mon, 13 Dec 2010 15:35:16 +0000 (15:35 +0000)]
Fix malformed specification XML.

bf532d6 and 28a2d08 each introduced errors in the specification XML that
left it malformed and unbuildable.

13 years agoBe more verbose when publishing or requesting session bus address on Windows.
Ralf Habacker [Tue, 7 Dec 2010 01:32:36 +0000 (02:32 +0100)]
Be more verbose when publishing or requesting session bus address on Windows.

13 years agoAdd documentation for autolaunch meta transport on Windows.
Ralf Habacker [Tue, 7 Dec 2010 01:23:39 +0000 (02:23 +0100)]
Add documentation for autolaunch meta transport on Windows.

Reviewed-by: Romain Pokrzywka
13 years agoAdd launchd section to specification.
Mike McQuaid [Wed, 8 Dec 2010 22:49:30 +0000 (22:49 +0000)]
Add launchd section to specification.

13 years agoImplements user limited session bus on Windows.
Ralf Habacker [Tue, 7 Dec 2010 01:16:14 +0000 (02:16 +0100)]
Implements user limited session bus on Windows.

This feature is enabled by using the session bus address
"autolaunch:scope=*user" for the client and the server.

Reviewed-by: Romain Pokrzywka
13 years agoFixed bug not freeing mutex handle on Windows.
Ralf Habacker [Tue, 7 Dec 2010 00:34:38 +0000 (01:34 +0100)]
Fixed bug not freeing mutex handle on Windows.

Reviewed-by: Romain Pokrzywka
13 years agoVariables and functions name cleanup on Windows.
Ralf Habacker [Tue, 7 Dec 2010 00:07:05 +0000 (01:07 +0100)]
Variables and functions name cleanup on Windows.

Reviewed-by: Romain Pokrzywka
13 years ago10.4 is old so set more sensible launchd defaults.
Mike McQuaid [Wed, 8 Dec 2010 22:58:32 +0000 (22:58 +0000)]
10.4 is old so set more sensible launchd defaults.

13 years agoFix bus-test failure with cmake on unix.
Ralf Habacker [Sun, 11 Apr 2010 09:46:04 +0000 (11:46 +0200)]
Fix bus-test failure with cmake on unix.

SHA1 connection tests needs an non debug-pipe transport for the activation bus.

13 years agoEnable launchd.
Benjamin Reed [Sun, 19 Jul 2009 15:45:34 +0000 (11:45 -0400)]
Enable launchd.

This patch enables support for Mac OS X's launch daemon
for startup as well as sharing of the DBus session bus
environment.  It includes a LaunchAgent plist for automatic
start of the session bus.

13 years agoLook up DISPLAY from launchd if not initialized.
Benjamin Reed [Tue, 21 Jul 2009 14:33:32 +0000 (10:33 -0400)]
Look up DISPLAY from launchd if not initialized.

13 years agoAdd launchd implementation.
Benjamin Reed [Sun, 19 Jul 2009 15:44:50 +0000 (11:44 -0400)]
Add launchd implementation.