| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
To allow dbus-glib to be built with G_DISABLE_ASSERT defined but tests
enabled (--disable-asserts --enable-tests), we need to explicitly
undefine it when building the tests, otherwise g_test_init() turns
into an abort since GLib 2.58.
We no longer actually use g_assert() for test-only code, so we can
remove the warning about tests' failing results not being reported.
test-profile.c still leaves assertions disabled: it does not use
g_test_init(), and it is intended to be used for profiling, where the
overhead of assertions is potentially significant.
Resolves: #16
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
|
|
|
| |
This avoids the assertions being removed by G_DISABLE_ASSERT, and gives
somewhat better diagnostic messages if the assertion fails.
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
|
|
| |
This is considered incorrect and causes a libdbus check failure.
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
|
|
|
|
| |
This replaces the inappropriate org.freedesktop.DBus.GLib.ErrorError
with the intended error codes, affecting two tests.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Resolves: https://gitlab.freedesktop.org/dbus/dbus-glib/issues/1
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
|
|
|
| |
Use dbus-run-session to run the only one that can be used as an
automated test right now.
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
|
|
|
|
| |
If we embed dbus-gmain in dbus-glib, dbus-python, at-spi2-core and
others as a submodule or subtree, we won't want it to export the
same ABI in all of them.
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
| |
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
|
|
| |
This will let us do a few tests with just dbus-gmain.
Signed-off-by: Simon McVittie <smcv@debian.org>
|
|
|
|
|
|
|
|
|
|
| |
The strange use of source from a sibling directory seems to be breaking
distcheck:
Making distclean in manual
make[3]: Entering directory '/.../dbus-glib-0.105/_build/sub/test/manual'
Makefile:409: ../core/.deps/my-object.Po: No such file or directory
make[3]: *** No rule to make target '../core/.deps/my-object.Po'. Stop.
|
|
|
|
|
| |
They haven't been needed since GLib 2.30. Until now, we were also
incorrectly distributing the generated marshallers in tarball releases.
|
|
|
|
| |
This now raises a g_warning() which makes the test fail.
|
| |
|
| |
|
|
|
|
| |
Since GLib 2.32, which we now depend on, this is unnecessary.
|
|
|
|
|
|
|
|
|
|
|
|
| |
This means we can assume that GLib and libdbus are thread-safe by
default.
Also explicitly document that the object-mapping layer of dbus-glib
is not thread-safe.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=64214
Reviewed-by: Colin Walters <walters@verbum.org>
[with review comments to be fixed in subsequent commit -smcv]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When switching a connection from one main loop to another, timeouts are
moved to the new connection, so connection_setup_add_timeout gets called on
the new connection for all timeouts that were set on the old connection.
This means that the timeout's data can, in fact, be non-NULL, and it will be
freed / removed from the old connection as a side-effect of adding it to the
new connection. A similar assert was removed from dbus_connection_add_watch
a while ago as part of the original patch for this bug.
Also, clean up the regression test a bit, and test that a DBusConnection
can be migrated when a pending call is present.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=30574
[resolve conflicts -smcv]
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
With modern GLib, misusing GSource provokes a critical warning, which
is made fatal by our test framework.
It is possible (indeed likely) that some or all of the
cancel_exit_timeout() calls just before re-adding the timeout are
redundant, but I didn't want to waste time on assessing that - easier
to just do it every time.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=83530
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
|
|
| |
When valgrinding tests it's useful to avoid using the shared bus, which
cannot be closed. Move all the essential shared-bus use into a test which
can be treated specially.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
| |
This frees more of the memory.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
| |
This frees more of the memory.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, if we'd freed the GValues, it would have crashed, because
g_value_take_boxed on a value of type G_TYPE_VALUE requires that the inner
GValue was g_malloc'd individually, but we were allocating them as a
block. This only "worked" because *none* of them were freed...
The returned string was also leaked; free it too, but only on success.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=41129
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
|
| |
Inspired by Sebastian Krahmer's 'darklena.c', which used
this bug to exploit pam_fprintd.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=68603
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
| |
The test for errors uses dbus_set_error() so it should link -ldbus-1.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=68601
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
| |
[rebased and added to run-test.sh -dbcw]
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=23633
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We close the connection to ourselves, resulting in libdbus shooting us
in the head unless we ask it not to. Nice to see how thoroughly this
code has been tested...
For future reference, it can be tested like this:
DBUS_TOP_BUILDDIR=$(pwd) ./test/core/run-test.sh profile
(or substitute the real absolute top build directory, if out-of-tree).
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=40711
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Dan Williams <dcbw@redhat.com>
|
|
|
|
|
|
|
|
|
| |
It's alarming how often the phrase "I'm amazed this ever worked" comes
up while maintaining dbus-glib.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=40711
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Dan Williams <dcbw@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
test-profile.c was its only use, and all it was used for was to listen
on an abstract or path-based socket, matching what D-Bus would do,
when obtaining comparative performance figures for D-Bus vs.
plain Unix sockets.
test-profile.c isn't normally run, only works on Unix, and
the check for abstract sockets was broken on current glibc (fd.o #29895),
leading to us using the non-abstract code path anyway, so this clearly
wasn't very important. I'm tempted to delete test-profile.c entirely,
but until then, let's just make it use path-based sockets.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=40711
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Dan Williams <dcbw@redhat.com>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=40711
Reviewed-by: Dan Williams <dcbw@redhat.com>
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=40711
Reviewed-by: Dan Williams <dcbw@redhat.com>
|