Closures use a 16-bit atomic reference count, which is really slow
on certain ARM64 CPUs such as the Cortex-A57 (glib#1316). This is
non-trivial to solve, since the public struct field cannot be enlarged
to 32-bit while preserving ABI, and 16-bit atomic operations would be new
(and rather niche) API.
Until this can be solved properly (hopefully in GLib 2.59.x), cut down
the number of signal emission cycles and bump up the timeout in the
Meson build system, so that builds won't time out. We can't just take
another zero off the number of signal emission cycles, as was done in the
original version of this patch in Debian, because if we do that it can
result in test failures when the main thread starves the other threads.
ARM64 CPUs are backwards-compatible with 32-bit ARM, and the same
slowdown can be seen when building and testing 32-bit code on these
CPUs, so check for both 32- and 64-bit ARM.
Bug-Debian: https://bugs.debian.org/880883
Co-authored-by: Iain Lane <laney@debian.org>
Signed-off-by: Simon McVittie <smcv@debian.org>
This makes it easier to debug test failures, by ensuring that g_debug()
and g_test_message() are printed as TAP diagnostics.
Fixes: https://gitlab.gnome.org/GNOME/glib/issues/1528
Signed-off-by: Simon McVittie <smcv@collabora.com>
These keep on taking just longer than 30s on my local machine when run
in parallel with the rest of the tests (i.e. with `ninja test`). Testing
them individually, they do terminate correctly.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Meson has the ability to classify tests according to "suites", a list of
tags. This is especially useful when we want to run specific sets of
tests — e.g. only GLib's tests — instead of the whole test suite. It
also allows us to classify special tests, like "slow" ones, so that we
can only run them when needed.
It's unnecessary, and only adds visual noise; we have been fairly
inconsistent in the past, but the semi-colon-less version clearly
dominates in the code base.
https://bugzilla.gnome.org/show_bug.cgi?id=669355
Recent changes to the properties testcase made invalid use of the GArray
free function. This free function takes a pointer to the item to be
freed, not the item itself. Since that item was a pointer to a GObject,
g_object_unref() was getting a GObject**, rather than a GObject*.
The use of GArray in this testcase is pretty questionable in the first
place, so just use C arrays instead.
Perform a substantial cleanup of the build system with respect to
building and installing testcases.
First, Makefile.decl has been renamed glib.mk and substantially
expanded. We intend to add more stuff here in the future, like canned
rules for mkenums, marshallers, resources, etc.
By default, tests are no longer compiled as part of 'make'. They will
be built when 'make check' is run. The old behaviour can be obtained
with --enable-always-build-tests.
--disable-modular-tests is gone (because tests are no longer built by
default). There is no longer any way to cause 'make check' to be a
no-op, but that's not very useful anyway.
A new glibtests.m4 file is introduced. Along with glib.mk, this
provides for consistent handling of --enable-installed-tests and
--enable-always-build-tests (mentioned above).
Port our various test-installing Makefiles to the new framework.
This patch substantially improves the situation in the toplevel tests/
directory. Things are now somewhat under control there. There were
some tests being built that weren't even being run and we run those now.
The long-running GObject performance tests in this directory have been
removed from 'make check' because they take too long.
As an experiment, 'make check' now runs the testcases on win32 builds,
by default. We can't run them under gtester (since it uses a pipe to
communicate with the subprocess) so just toss them in TESTS. Most of
them are passing on win32.
Things are not quite done here, but this patch is already a substantial
improvement. More to come.
Very many testcases, some GLib tools (resource compiler, etc) and
GApplication were calling g_type_init().
Remove those uses, as they are no longer required.
https://bugzilla.gnome.org/show_bug.cgi?id=686161
Either g_type_register_static_simple (used by G_DEFINE_TYPE_EXTENDED)
and G_IMPLEMENT_INTERFACE use automatic variables for GTypeInfo and
GInterfaceInfo structs, while tutorials and source code often use
static variables. This commit consistently adopts the former method.
https://bugzilla.gnome.org/show_bug.cgi?id=600161
2007-11-21 21:06:47 Tim Janik <timj@imendio.com>
* Makefile.decl: initialize automake variables EXTRA_DIST and
TEST_PROGS for unconditional appending via += in other makefiles.
define recursive test targets: test, test-report, perf-report,
full-report, as described here:
http://mail.gnome.org/archives/gtk-devel-list/2007-November/msg00000.html
* Makefile.am:
* build/win32/vs8/Makefile.am, build/win32/dirent/Makefile.am:
* build/win32/Makefile.am, build/Makefile.am:
* docs/Makefile.am, docs/reference/Makefile.am:
* docs/reference/glib/Makefile.am, docs/reference/gobject/Makefile.am:
* gmodule/Makefile.am, tests/Makefile.am:
* tests/refcount/Makefile.am, tests/gobject/Makefile.am:
* glib/update-pcre/Makefile.am, glib/libcharset/Makefile.am:
* glib/tests/Makefile.am, glib/pcre/Makefile.am:
* glib/gnulib/Makefile.am, gobject/Makefile.am, m4macros/Makefile.am:
* gthread/Makefile.am, glib/Makefile.am:
include $(top_srcdir)/Makefile.decl, adapted EXTRA_DIST assignments.
* glib/tests/Makefile.am: removed example testing rules.
* glib/tests/testing.c: conditionalized performance and slow tests.
* glib/gtestutils.h:
* glib/gtestutils.c: work around g_test_config_vars not changing its
exported value after value assignments, aparently due to symbol aliases.
* glib/gtester.c: fixed off-by-one error which produced junk in logs.
* configure.in: check for python >= 2.4 and provide $PYTHON for scripts.
svn path=/trunk/; revision=5914
* refcount/signals.c:
* refcount/objects.c:
* refcount/objects2.c:
* refcount/closures.c:
* refcount/properties.c:
* refcount/properties2.c: changed namespace prefix from g_test_* to my_test_*
to not clash with newly introduced g_test* API in glib.
svn path=/trunk/; revision=5875
2007-07-09 Loïc Minier <lool@dooz.org>
* tests/refcount/closures.c: (main): Output newlines after thousand
iterations of the inner-loop of the closures test; this helps having
smaller lines and continuously outputting new lines. (#447048).
svn path=/trunk/; revision=5609
2005-09-11 Sebastian Wilhelmi <seppi@seppi.de>
* tests/refcount/Makefile.am (INCLUDES): Link the the refcount
tests to the system thread library $(G_THREAD_LIBS). Fixes#313744
and #314217.
2005-08-02 Tor Lillqvist <tml@novell.com>
* tests/refcount/objects.c
* tests/refcount/properties.c
* tests/refcount/signals.c: Use g_usleep() instead of sleep() for
portability.
Mon Aug 1 23:33:47 2005 Tim Janik <timj@imendio.com>
* tests/refcount/closures.c: test high contention on closure
reference counts to trigger and catch non-atomic updates.
* tests/refcount/objects.c:
* tests/refcount/objects2.c:
* tests/refcount/properties.c:
* tests/refcount/properties2.c:
* tests/refcount/signals.c:
fixed up test and threading fundamentals. variables accessed from all
threads need to be volatile. context switches are enforced by using
g_thread_yield(), not g_usleep(1) which may result in busy waits on
some platforms. for testcode, always consider all warnings and
critical messages fatal. issue the currently running program on
stdout. improved progress indicators.
* tests/refcount/properties.c:
* tests/refcount/objects.c:
don't overdo the number of testing threads to keep the testing machine
usable, 2 threads can produce as much contention as 20 if executing the
same code.
* tests/refcount/signals.c: only start 1 thread per object. GObject
doesn't provide mutually exclusive object access, but only mutually
exclusive reference count modification.
* tests/Makefile.am: added closures test.
2005-07-15 Matthias Clasen <mclasen@redhat.com>
Make refcounting threadsafe by using atomic
operations. (#166020, Wim Taymans)
* gobject.c: Use a recursive lock to protect the
notify queue.
(g_object_unref): Get rid of g_object_last_unref and
do the last unref handling in g_object_unref.
(g_object_ref, g_object_unref): Use atomic operations.
* gsignal.c (struct _HandlerMatch): Use a full integer
for the ref_count field.
(handler_ref, handler_unref_R): Use atomic operations.
* gparam.c (g_param_spec_ref, g_param_spec_unref):
Use atomic operations instead of a lock to make the
refcounting threadsafe.
* gclosure.c (g_closure_ref, g_closure_unref): Use atomic
operations. This is more complicated here, since the
refcount is stored in a bitfield, so we also have
to access all other bitfield members atomically.
* gsignal.c (handlers_find): Read the meta_marshal flag
of the closure atomically.
* tests/Makefile.am (SUBDIRS): Add tests/refcount
* configure.in: Add tests/refcount
* tests/refcount/properties.c: Test property changes
from multiple threads.
* tests/refcount/signals.c: Test signal emission from
multiple threads.
* tests/refcount/objects.c: Test refcounting from
multiple threads.
* tests/refcount/objects2.c:
* tests/refcount/properties2.c: Tests to measure the
overhead of threadsafe refcounting.
* glib/giochannel.c (g_io_channel_ref, g_io_channel_unref):
Use atomic operations to make refcounting
threadsafe. (#166020, Wim Taymans)
