Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.

BUG=640800
Committed: https://crrev.com/4889f1867a957f6dd401ce885e6c56ff864b1a4c
Cr-Commit-Position: refs/heads/master@{#414285}

[watk, 2016-08-24 22:27:58]

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

[watk, 2016-08-24 22:28:18]

Description was changed from

==========
Invalidate VideoRendererImpls weak ptrs before reseting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.
==========

to

==========
Invalidate VideoRendererImpls weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.
==========

[watk, 2016-08-24 22:28:28]

Description was changed from

==========
Invalidate VideoRendererImpls weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.
==========

to

==========
Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.
==========

[DaleCurtis, 2016-08-24 22:33:27]

lgtm; if it really does improve things we might see seek times improve, so it
may be worth attaching this to a bug for merging later.

[DaleCurtis, 2016-08-24 22:33:38]

(See seek times improve in telemetry that is)

[watk, 2016-08-24 22:52:52]

Description was changed from

==========
Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.
==========

to

==========
Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.

BUG=640800
==========

[watk, 2016-08-24 22:54:37]

On 2016/08/24 22:33:38, DaleCurtis wrote:
> (See seek times improve in telemetry that is)

Thanks, made a bug anyway. I didn't see seek times change in a local telemetry
run even though AVDA::ReusePictureBuffer definitely does MaybeRenderEarly and
DoIOTask(). Maybe we don't have that many outstanding frames with AVDA so the
impact isn't 4x ReusePictureBuffer but only 1 or 2.

[watk, 2016-08-24 22:54:50]

The CQ bit was checked by watk@chromium.org

[commit-bot: I haz the power, 2016-08-24 22:55:46]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-08-25 01:00:03]

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

[commit-bot: I haz the power, 2016-08-25 01:00:04]

Try jobs failed on following builders:
  chromeos_daisy_chromium_compile_only_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromeos_daisy_...)
  linux_chromium_clobber_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[watk, 2016-08-25 01:15:16]

The CQ bit was checked by watk@chromium.org

[commit-bot: I haz the power, 2016-08-25 01:16:28]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-08-25 03:16:34]

Description was changed from

==========
Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.

BUG=640800
==========

to

==========
Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.

BUG=640800
==========

[commit-bot: I haz the power, 2016-08-25 03:16:35]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-08-25 03:20:52]

Description was changed from

==========
Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.

BUG=640800
==========

to

==========
Invalidate VideoRendererImpl's weak ptrs before resetting the decoder

Previously VideoRendererImpl waited for the VideoFrameStream::Reset() to
complete before invalidating its frame callback weak pointers, which
meant that it could receive frames that it didn't want between Flush()
and OnVideoFrameStreamResetDone(). For simplicity, now the weak pointers
are invalidated at the same time that Reset() is called, so it's
guaranteed that no frames will be received after Flush().

Also included in this change is reordering the deletion of the video
frame queue and resetting of the VideoFrameStream to save VDAs doing
wasted work.

BUG=640800

Committed: https://crrev.com/4889f1867a957f6dd401ce885e6c56ff864b1a4c
Cr-Commit-Position: refs/heads/master@{#414285}
==========

[commit-bot: I haz the power, 2016-08-25 03:20:53]

Patchset 1 (id:??) landed as
https://crrev.com/4889f1867a957f6dd401ce885e6c56ff864b1a4c
Cr-Commit-Position: refs/heads/master@{#414285}