We had problems because we'd call Destroy on a port after calling Destroy on
its streams. This patch makes the port's Destroy ControlMessage not use a stream,
instead we get the stream from the port directly. It also makes us update the
graph's mPortCount only when the port is finally destroyed; this fixes a potential
bug where the current graph could shut down when all streams are removed, before
all ports have been completely destroyed.
--HG--
extra : rebase_source : 3245fa64b79fb8a75c17190f1a4a2af6d42a5012
Until now we've identified the "first active track" of a stream and played only that. Instead,
it makes more sense to play all the tracks.
For video, we pick the last track that has a video frame for the current time, and display that.
--HG--
extra : rebase_source : 00ce3eb363df06c292232aa507e861639d10cff2
The main thing this patch does is to add ProcessedMediaStream objects and
MediaInputPorts connecting streams. ProcessedMediaStreams are an abstract
class that doesn't constrain what processing is actually performed, except
that for now we assume a stream's processing depends only on its inputs
window of data between mCurrentTime and mStateComputedTime.
This patch reorganizes the way the blocking status of each stream is computed.
The streams are partitioned into groups so that every stream which can affect
the blocking status of another stream is in the same group as that other stream.
We also add a pass to order the streams by dependency so we can process the streams
in order of dependency; this pass also identifies the streams that form part of a
cycle.
--HG--
extra : rebase_source : c45c931a264e73f295642a934500bbeaa6448774
This is actually a better fix for bug 774597. We can rely on main thread code to call
Destroy on all the MediaStreams normally. These Destroy calls can be handled even when the
MediaGraph thread has been shut down, since MediaStreamGraphImpl::AppendMessage will
call RunDuringShutdown on the Destroy messages.
Rename mBlockingDecisionsMadeUntilTime to mStateComputedTime. The invariant is that
all graph state is known up to mStateComputedTime but not beyond it (except for some
stream contents that may be buffered beyond it).
Get rid of mMessageAffectedTime and all the code around computing "affected times" and "action times".
Instead, all messages take effect at mStateComputedTime.
Get rid of the two-phase execution of messages. Everything can just happen in a single Run() method.
The first part just handles the case where nsAudioStream failed to allocate a stream. It won't be playing
anything, so instead of trying to get the audio position, just fall back to the media graph current time.
Otherwise GetPositionInFrames returns -1 and things go badly from there.
The second part simplifies the calculation of the next mCurrentTime to just make it based on real time.
We had some code to not let it advance past the end of a stream's buffer, but the next part will make that
unnecessary.
The third part is the real fix. When the new current time has advanced past mBlockingDecisionsMadeUntilTime,
that means the control loop didn't run in time to replenish the audio output buffers and keep up with its
other duties. Effectively all streams have been blocked between mBlockingDecisionsMadeUntilTime and
the new current time. Account for that by adding the difference as extra blocked time for every stream.
We only need to ensure that the stream is marked blocked from mBlockingDecisionsMadeUntilTime indefinitely
far into the future, and then update mBlockingDecisionsMadeUntilTime to the new current time, because the
code takes into account that only blocking decisions up to mBlockingDecisionsMadeUntilTime are valid.
