CI now checks that plugin caches are up to date, for example
when adding a new property to an element.
This is something pretty easy to forget, and it can be checked
in a pre-commit hook provided the cache generator generates no
unnecessart diff on the developer's machine. This is now the
case for me in core, -base and -good, and eventually all the
repositories should behave appropriately, at least on my
development machine.
The new check in the pre-commit hook is only performed when
the user is in the development environment (it checks
$MESON_BUILD_ROOT to determine that), and when the developer
has opted-in by setting `GST_CACHE_HOOK=enabled`. That is
because the hook will actually rebuild the cache with ninja,
and modify it in the source directory, that's not a behaviour
we want to enable by default.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/546>
There is a race-condition that can trigger the assertion in
gst_bus_add_signal_watch_full():
If gst_bus_add_signal_watch_full() is called immediately after
gst_bus_remove_signal_watch() then bus->priv->signal_watch may still be set
because gst_bus_source_dispose() or gst_bus_source_finalize() was not yet
called.
This happens if the corresponding GMainContext has the source queued for
dispatch. In this case, the following dispatch will only unref and delete
the signal_watch because it was already destroyed. Any pending messages
will remain until a new watch is installed.
So bus->priv->signal_watch can be cleared immediately when the watch is
removed. This avoid the race condition.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/543>
Instead do everything it did as part of GObject::constructed() and
change the function to always return TRUE.
gst_ghost_pad_construct() was meant to be called by subclasses right
after construction of the object to finish construction as it can fail
in theory. In practice it's impossible for it to fail, even more so if
called directly from GObject::constructed(): The only failure condition
is if the newly created proxy pad already has a parent, which is
impossible at this point as nothing else can have a reference to it.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/540>
Since glib 2.64, gthreadpool will start waiting on a GCond immediately upon
creation. This can cause issues if we fork *before* actually using the
threadpool since we will then be signalling that GCond ... from another process
and that will never work.
Instead, delay creationg of thread pools until the very first time we need
them. This introduces a minor (un-noticeable) delay when needing a new thread
but fixes the issues for all users of GSTreamer that will call gst_init, then
fork and actually start pipelines.
See https://gitlab.gnome.org/GNOME/glib/-/issues/2131 for more context.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/531>
The signal handlers were performing mutex operations in the signal handlers
which is bad idea that may lead to deadlocks.
1. Implement a separate signal thread to handle the signals.
2. Use the glib provided signal GSource to avoid performing operations in
the signal handler.
Fix#186
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/487>
When compiling for 32bit ios arm, the static assert that the
GstClockEntryImpl smaller or equal to the struct _GstClockEntryImpl
triggered. (they were 12bytes off).
To fix this, the padding is increased by 12 bytes (on 32bit).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/525>
These can be passed to gst_type_mark_as_plugin_api, to inform
plugin cache generation.
For now a single flag is specified, "IGNORE_ENUM_MEMBERS", it
can be used for dynamically generated enums to avoid documenting
environment-specific enumeration members. An example is
GstX265EncTune.
Since those are using the clock for sync, they need to also
provide a clock for good measure. The reason is that even if
downstream elements provide a clock, we don't want to have
that clock selected because it might not be running yet.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/509>
When we want to perform a downstream bitrate query, just
set the reconfigure flag on the srcpad and get the streaming
thread to do it. That avoids emitting a downstream query
when receiving the upstream RECONFIGURE event - which can
lead to deadlocks if downstream is sending the event from
within a lock - e.g. input-selector.
If querying the downstream bitrate changes the cached
value, then make sure to update our buffering state
and potentially post a BUFFERING message to the application.
Fixes: https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/566
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/501>
The deadlock was the following:
* One thread requests a new pad, the internal lock is kept while adding the pad
* Another thread (or the same one) requests the internal links of a pad (could
be that pad)... which also requires that lock.
That internal lock is not required when adding the pad to the element (which is
the last action when requesting a new pad). The fact it will be actually used
will be *after* the request pad function is released.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/512>
