Increase default drift tolerance to 40ms to avoid glitches with decoders
or formats where there's a lot of timestamp jitter for some reason or
another (in this case: asf/wma), at least until we implement timestamp
smoothing.
Our calibration against the pipeline clock is done with the adjusted
ringbuffer time, so take the adjustement into account. Fixes some audio dropouts
when reusing audio sinks after switching clocks and slaving methods in a
pipeline.
When we are calibrating the internal clock against the external clock take into
account the time offset applied to our internal clock because we will subtract
that in the render_function again.
Add drift-tolerance property (defaulting to 20ms) to handle resync after clock
drift or timestamp drift instead of relying on the latency-time value for clock
drift and 500ms for timestamp drift.
Remove warning about discont timestamp and simply resync. The warning is in some
cases not correct and is triggered more frequently now that we lower the
tolerance value.
There's not much point in using GST_DEBUG_FUNCPTR with GObject
virtual functions such as get_property, set_propery, finalize and
dispose, since they'll never be used by anyone anyway. Saves a
few bytes and possibly a sixteenth of a polar bear.
Check for pulsesink < 0.10.17 because it includes code that is now included in
baseaudiosink. Disable that code in baseaudiosink to be compatible with the
older version.
Take the time of the clock so that the last_time field is set. This is important
for sinks that restart their internal ringbuffer after a caps change and need to
know the last know position.
When going to NULL, we reset the ringbuffer so that it starts beck from 0. We
also make sure that the clock is updated with the elapsed time so that it
alsways increments even when the ringbuffer goes back to 0. When this happened
we need to adjust the sample position for the reset ringbuffer.
Fixes#594136
Unparent and free the ringbuffer when going to NULL, like we do with the
audiosrc element. We can do this now because we correctly manage the time
jumping back to 0.
Use the unadjusted internal clock times to calculate the internal/external
offset when calibrating the clock.
When going to NULL, unparent and free the ringbuffer, like we do in the source
element.
See #578506
After we pause the stream and interrupt the writeout to the ringbuffer, also adjust
the amount of output samples we consumed. We can't do this reliably with the
current API when we are doing trick modes but we can do the right thing for
normal playback.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_change_state):
Avoid holding the OBJECT_LOCK when calling ringbuffer functions that
take the ringbuffer lock because rinbuffer lock > OBJECT_LOCK. We can do
this because the async_play method is deprecated and usually not called
anymore.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_setcaps):
Pause the write thread before deactivating and releasing the ringbuffer
to avoid a deadlock when we do gapless playback with different sample
rates in playbin2. Fixes#564929.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_slave_method_get_type),
(gst_base_audio_sink_class_init):
* gst-libs/gst/audio/gstbaseaudiosink.h:
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_slave_method_get_type),
(gst_base_audio_src_class_init):
* gst-libs/gst/audio/gstbaseaudiosrc.h:
API: Add GST_TYPE_BASE_AUDIO_(SRC|SINK)_SLAVE_METHOD to the
public API. This is needed for the C++ bindings to be able
to use this base classes. Fixes bug #564200, #564206.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_drain), (gst_base_audio_sink_event),
(gst_base_audio_sink_sync_latency), (gst_base_audio_sink_render),
(gst_base_audio_sink_change_state):
Really fix audiosink drain handling by keeping track of the running_time
of the last sample.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
Time is already in running_time. Remove base_time handling. Fixes
audiosinks not draining and thus chopping some audio in the end.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
Add one log message to check for audio_drained. Sync one log message
with the condition. Send EOS after draining audio in pull mode.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_sync_latency), (gst_base_audio_sink_render),
(gst_base_audio_sink_callback):
Use gst_base_sink_do_preroll() to wait for PLAYING and before waiting
for the latency to expire, fixes#559567.
Original commit message from CVS:
* gst-libs/gst/audio/gstaudiosink.c:
(gst_audioringbuffer_class_init), (gst_audioringbuffer_acquire),
(gst_audioringbuffer_activate), (gst_audioringbuffer_release),
(gst_audioringbuffer_stop):
Implement a separate activate functions to start monitoring the segments
or, in pull mode, pulling in data.
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_init), (gst_base_audio_sink_dispose),
(gst_base_audio_sink_query_pad), (gst_base_audio_sink_query),
(gst_base_audio_sink_setcaps), (gst_base_audio_sink_callback),
(gst_base_audio_sink_activate_pull),
(gst_base_audio_sink_async_play),
(gst_base_audio_sink_change_state):
Implement pad and element convert query function.
Activate the ringbuffer.
Use the segment last_stop value as the offset to pull.
Use new basesink _do_preroll() method to preroll in the pulling thread.
Take appropriate locking in the pulling thread.
* gst-libs/gst/audio/gstringbuffer.h:
Update some docs.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_query), (gst_base_audio_sink_skew_slaving),
(gst_base_audio_sink_render):
Report latency even if we are not live instead of hiding it.
Take ts-offset and render-delay of the basesink into account when
scheduling samples.
Rework the clipping code so that we can take the various offsets into
account and still do correct clipping.
Original commit message from CVS:
* gst-libs/gst/audio/gstaudioclock.c: (gst_audio_clock_init),
(gst_audio_clock_reset), (gst_audio_clock_get_internal_time):
* gst-libs/gst/audio/gstaudioclock.h:
Add method to inform the clock that the time starts from 0 again. We use
this info to calculate a clock offset so that the time we report in
internal_time is monotonically increasing, as required by the clock base
class. Fixes#521761.
API: GstAudioClock::gst_audio_clock_reset()
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_skew_slaving),
(gst_base_audio_sink_change_state):
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_create), (gst_base_audio_src_change_state):
Reset reported time when we (re)create the ringbuffer.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_sync_latency):
We can only use our optimal calibration if we prerolled before the
latency expired.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_resample_slaving),
(gst_base_audio_sink_skew_slaving),
(gst_base_audio_sink_sync_latency), (gst_base_audio_sink_render),
(gst_base_audio_sink_async_play),
(gst_base_audio_sink_change_state):
Change the way in which the ringbuffer is started when dealing with a
slaved clock and latency. We now sync to the clock until we reach
upstream latency before starting the ringbuffer. This has the effect
that we can accurately align the master and slave clocks and let the
rate correction code take care of the initial drift or rounding errors
instead of leaving them uncorrected with the old approach.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_render), (gst_base_audio_sink_async_play):
Revert previous patch that attempted to more accurately calculate the
initial offset between master and slave clock. The best thing we can do
in general is take the time of both clocks as the diff since we don't
know when the actual preroll happened.
Original commit message from CVS:
* gst-libs/gst/audio/gstaudiosink.c: (gst_audioringbuffer_acquire):
Choose to allocate one less segment but require one additional segment
as latency.
* gst-libs/gst/audio/gstaudiosrc.c: (gst_audioringbuffer_acquire):
No need to increment the number of segments in the source.
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_get_time), (clock_convert_external),
(gst_base_audio_sink_resample_slaving),
(gst_base_audio_sink_skew_slaving),
(gst_base_audio_sink_none_slaving), (gst_base_audio_sink_render),
(gst_base_audio_sink_async_play):
Remove adding latency when returning the internal time while subtracting
it again when we use the value a little later.
When calculating the end timestamp, we are making a rounding error
with the current algorithm. Ensure that we don't accumulate these
rounding errors when aligning samples by not resampling at all if we
don't need to. Fixes#419351.
Make the initial calibration of the clock slaving a little more
predictable and accurate. Also handle the case where we don't do
clock slaving.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_query):
Report the latency with the new seglatency parameter.
* gst-libs/gst/audio/gstringbuffer.c:
(gst_ring_buffer_debug_spec_buff), (gst_ring_buffer_parse_caps),
(gst_ring_buffer_acquire):
* gst-libs/gst/audio/gstringbuffer.h:
Add new field to the ringbufferspec to specify the expected latency
between the underlying device read/write pointer, this is needed
when writing sinks that sit a little closer to the hardware.
Add some more docs for other fields.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_class_init):
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_class_init):
* gst/playback/gstplayback.c: (plugin_init):
* gst/volume/gstvolume.c: (plugin_init):
Work around missing bits of thread-safety on older GLibs some
more to avoid assertions when starting up multiple playbin
objects concurrently (see #512382).
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_skew_slaving), (gst_base_audio_sink_render):
Guard against over and underflows because of clock slaving.
When we are using our own clock, still compensate for any calibrations
that we might have done to our clock.
Original commit message from CVS:
* gst-libs/gst/audio/gstaudiosink.c: (gst_audioringbuffer_release):
Small debug improvement.
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_render):
Fix bug in determining the sample start/stop position, we want to base
this decision on the fact that we are going forwards or backwards, not
slower or faster. This fixes some ugly resync warnings when playing at
very slow speeds.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_class_init):
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_class_init):
Ref audio clock class from a thread-safe context to make sure
we're not bit by GObjects lack of thread-safety here (#349410),
however unlikely that may be in practice.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_drain):
Our EOS time contains the base_time, _wait_eos() expects a running_time
so we have to subtract the base_time again before calling the function.
This fixes an EOS regression where the base_time was added twice and EOS
took longer and longer in certain situations.
Fixes#498767.
Original commit message from CVS:
* docs/libs/gst-plugins-base-libs-sections.txt:
* gst-libs/gst/audio/gstbaseaudiosink.c: (slave_method_get_type),
(gst_base_audio_sink_set_provide_clock),
(gst_base_audio_sink_get_provide_clock),
(gst_base_audio_sink_set_slave_method),
(gst_base_audio_sink_get_slave_method),
(gst_base_audio_sink_set_property),
(gst_base_audio_sink_get_property), (gst_base_audio_sink_drain),
(gst_base_audio_sink_none_slaving),
(gst_base_audio_sink_handle_slaving):
* gst-libs/gst/audio/gstbaseaudiosink.h:
Expose methods for some object properties so that subclasses can more
easily configure them.
Added slave method none, that completely disables slaving to the
internal clock.
API: gst_base_audio_sink_set_provide_clock()
API: gst_base_audio_sink_get_provide_clock()
API: gst_base_audio_sink_set_slave_method()
API: gst_base_audio_sink_get_slave_method()
* gst-libs/gst/audio/gstbaseaudiosrc.c:
(gst_base_audio_src_set_provide_clock),
(gst_base_audio_src_get_provide_clock),
(gst_base_audio_src_set_property),
(gst_base_audio_src_get_property), (gst_base_audio_src_create):
* gst-libs/gst/audio/gstbaseaudiosrc.h:
Expose methods for some object properties so that subclasses can more
easily configure them.
API: gst_base_audio_src_set_provide_clock()
API: gst_base_audio_src_get_provide_clock()
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_drain), (gst_base_audio_sink_change_state):
Also explicitly release the ringbuffer when going to NULL because it
is required in the setcaps function, before the state change to PAUSED
completes.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_drain):
Use new basesink method to make our EOS drain interruptable.
Original commit message from CVS:
* gst-libs/gst/audio/gstbaseaudiosink.c:
(gst_base_audio_sink_async_play):
When slaved to the clock, don't try to align a sample with the previous
one when going to PLAYING again.
