Fix underflow and av/sync issues in low delay and stall presence.

Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=599908, 601066
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.

Committed: https://crrev.com/675f8b4aacfec7f7fb1d33f430996672c4dc3eec
Cr-Commit-Position: refs/heads/master@{#389564}

[DaleCurtis, 2016-04-09 00:54:51]

The CQ bit was checked by dalecurtis@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-04-09 00:55:00]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1863373002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1863373002/20001

[commit-bot: I haz the power, 2016-04-09 04:08:58]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-04-09 04:08:58]

Dry run: This issue passed the CQ dry run.

[DaleCurtis, 2016-04-11 20:51:08]

Description was changed from

==========
bla

BUG=
==========

to

==========
Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=527990, 599908
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.
==========

[DaleCurtis, 2016-04-11 20:51:08]

dalecurtis@chromium.org changed reviewers:
+ xhwang@chromium.org

[DaleCurtis, 2016-04-11 20:52:00]

Preliminary version of the underflow fix. Needs working tests.

[DaleCurtis, 2016-04-11 23:46:58]

Patchset #4 (id:60001) has been deleted

[DaleCurtis, 2016-04-11 23:50:57]

Okay this is all done now; test all teh things! I've confirmed the new tests
fail w/ the old code as well.

[DaleCurtis, 2016-04-12 00:54:58]

Description was changed from

==========
Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=527990, 599908
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.
==========

to

==========
Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=599908
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.
==========

[xhwang, 2016-04-14 21:28:00]

LGTM with tiny nit

https://chromiumcodereview.appspot.com/1863373002/diff/80001/media/renderers/...
File media/renderers/video_renderer_impl_unittest.cc (right):

https://chromiumcodereview.appspot.com/1863373002/diff/80001/media/renderers/...
media/renderers/video_renderer_impl_unittest.cc:282: enum class
UnderflowTestType { kNormal, kLowDelay, kCantReadWithoutStalling };
nit: use UPPER_CASE style for enums?

[DaleCurtis, 2016-04-15 01:29:10]

I hate upper case :) But will do it before landing. Going to hold off on this
until we sort out https://codereview.chromium.org/1892013002 otherwise when
videos are scrolled off screen they will underflow and stall out (which is the
intended behavior), but is worse than letting the unwatched video play audio
uninterrupted even when it falls far behind.

[DaleCurtis, 2016-04-25 19:06:06]

Description was changed from

==========
Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=599908
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.
==========

to

==========
Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=599908, 601066
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.
==========

[DaleCurtis, 2016-04-25 19:06:27]

Landing this now that the other CL has landed and I have a final fix for slow
playback issues at https://codereview.chromium.org/1921803002

[DaleCurtis, 2016-04-25 19:06:31]

The CQ bit was checked by dalecurtis@chromium.org

[DaleCurtis, 2016-04-25 19:06:31]

The patchset sent to the CQ was uploaded after l-g-t-m from xhwang@chromium.org
Link to the patchset: https://codereview.chromium.org/1863373002/#ps100001
(title: "Rebase. Rename enum.")

[commit-bot: I haz the power, 2016-04-25 19:07:02]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1863373002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1863373002/100001

[commit-bot: I haz the power, 2016-04-25 20:25:44]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-04-25 20:25:45]

Failed to commit the patch.

[commit-bot: I haz the power, 2016-04-25 20:25:54]

The CQ bit was unchecked by commit-bot@chromium.org

[DaleCurtis, 2016-04-25 21:23:41]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2016-04-25 21:24:13]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1863373002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1863373002/100001

[commit-bot: I haz the power, 2016-04-25 21:30:04]

Committed patchset #5 (id:100001)

[commit-bot: I haz the power, 2016-04-25 21:32:39]

Description was changed from

==========
Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=599908, 601066
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.
==========

to

==========
Fix underflow and av/sync issues in low delay and stall presence.

When in low delay mode we were previously transitioning to HAVE_ENOUGH
if any frame was available; this may include frames which are bad. I.e.
frames that are too far in the past to be effective. Instead we need to
enter the underflow state in these cases and wait for effective frames
to enter the queue.

In low delay this should just add a frame or two of latency while we
catchup since the frames should be readily available in most cases. In
the stalled case, i.e. VideoFrameStream::CanReadWithoutStalling() is
false, this will attempt to purge frames as fast as possible such that
the decoder can catch up.

This is done by taking advantage of a new VideoRendererAlgorithm::Reset()
call which allows the frame queue to be purged while maintaining some of
the next frame estimates.

In all cases this patch also adds the ability to try and avoid ever
reaching the paused underflow state by removing expired frames as they
are enqueued if the decoder is not keeping.

BUG=599908, 601066
TEST=new unittest, manual testing w/ AVDA adaptivePlayback=false.

Committed: https://crrev.com/675f8b4aacfec7f7fb1d33f430996672c4dc3eec
Cr-Commit-Position: refs/heads/master@{#389564}
==========

[commit-bot: I haz the power, 2016-04-25 21:32:39]

Patchset 5 (id:??) landed as
https://crrev.com/675f8b4aacfec7f7fb1d33f430996672c4dc3eec
Cr-Commit-Position: refs/heads/master@{#389564}