Move encoder thread to VideoSendStream.

Move encoder thread to VideoSendStream.

Makes VideoCaptureInput easier to test and enables running more things
outside VideoCaptureInput on the encoder thread in the future
(initializing encoders and reconfiguring them, for instance).

BUG=webrtc:5410, webrtc:5494
R=stefan@webrtc.org

Committed: https://crrev.com/a4c76882b99ef38e5f5ddd1d325de8cdca3cca02
Cr-Commit-Position: refs/heads/master@{#11860}

[pbos-webrtc, 2016-03-03 14:08:11]

PTAL, I think this is a lot cleaner than before (and makes future changes
waaaaaaaaaaaaaaaaaaaaaay cleaner).

[pbos-webrtc, 2016-03-03 14:08:19]

The CQ bit was checked by pbos@webrtc.org to run a CQ dry run

[commit-bot: I haz the power, 2016-03-03 14:08:28]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1763693002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1763693002/1

[the sun, 2016-03-03 14:16:46]

solenberg@webrtc.org changed reviewers:
+ solenberg@webrtc.org

[the sun, 2016-03-03 14:16:47]

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
File webrtc/video/video_capture_input.h (left):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
webrtc/video/video_capture_input.h:66: rtc::PlatformThread encoder_thread_;
Does this mean we're going from one thread per capturer, to one thread per video
send stream? We're seeing some potential problems due to using too many threads.
See: https://bugs.chromium.org/p/chromium/issues/detail?id=582441 - this is
another side of the coin though...

[stefan-webrtc, 2016-03-03 14:19:36]

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
File webrtc/video/video_capture_input_unittest.cc (right):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
webrtc/video/video_capture_input_unittest.cc:60: const_cast<const
VideoFrame*>(&frame)->buffer(kYPlane));
Possibly static_cast?

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_send_strea...
File webrtc/video/video_send_stream.cc (right):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_send_strea...
webrtc/video/video_send_stream.cc:371: if
(!encoder_wakeup_event_.Wait(kThreadWaitTimeMs))
kWaitForevah

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/vie_encoder.h
File webrtc/video/vie_encoder.h (right):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/vie_encoder.h#ne...
webrtc/video/vie_encoder.h:82: void EncodeVideoFrame(VideoFrame video_frame);
Should this be a const reference?

[pbos-webrtc, 2016-03-03 14:25:32]

PTAL

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
File webrtc/video/video_capture_input.h (left):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
webrtc/video/video_capture_input.h:66: rtc::PlatformThread encoder_thread_;
On 2016/03/03 14:16:46, the sun wrote:
> Does this mean we're going from one thread per capturer, to one thread per
video
> send stream? We're seeing some potential problems due to using too many
threads.
> See: https://bugs.chromium.org/p/chromium/issues/detail?id=582441 - this is
> another side of the coin though...

Nope, this VideoCaptureInput is already one per send stream. And we reasonably
have one per encoder for performance/reduced delay reasons.

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
File webrtc/video/video_capture_input_unittest.cc (right):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_capture_in...
webrtc/video/video_capture_input_unittest.cc:60: const_cast<const
VideoFrame*>(&frame)->buffer(kYPlane));
On 2016/03/03 14:19:35, stefan-webrtc (holmer) wrote:
> Possibly static_cast?

Done.

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_send_strea...
File webrtc/video/video_send_stream.cc (right):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/video_send_strea...
webrtc/video/video_send_stream.cc:371: if
(!encoder_wakeup_event_.Wait(kThreadWaitTimeMs))
On 2016/03/03 14:19:35, stefan-webrtc (holmer) wrote:
> kWaitForevah

Done.

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/vie_encoder.h
File webrtc/video/vie_encoder.h (right):

https://codereview.webrtc.org/1763693002/diff/1/webrtc/video/vie_encoder.h#ne...
webrtc/video/vie_encoder.h:82: void EncodeVideoFrame(VideoFrame video_frame);
On 2016/03/03 14:19:35, stefan-webrtc (holmer) wrote:
> Should this be a const reference?

Done.

[pbos-webrtc, 2016-03-03 14:25:42]

feedback

[stefan-webrtc, 2016-03-03 14:56:51]

lgtm

[pbos-webrtc, 2016-03-03 15:29:12]

Description was changed from

==========
Move encoder thread to VideoSendStream.

Makes VideoCaptureInput easier to test and enables running more things
outside VideoCaptureInput on the encoder thread in the future
(initializing encoders and reconfiguring them, for instance).

BUG=webrtc:5410, webrtc:5494
R=stefan@webrtc.org
==========

to

==========
Move encoder thread to VideoSendStream.

Makes VideoCaptureInput easier to test and enables running more things
outside VideoCaptureInput on the encoder thread in the future
(initializing encoders and reconfiguring them, for instance).

BUG=webrtc:5410, webrtc:5494
R=stefan@webrtc.org

Committed:
https://chromium.googlesource.com/external/webrtc/+/a4c76882b99ef38e5f5ddd1d3...
==========

[pbos-webrtc, 2016-03-03 15:29:14]

Committed patchset #2 (id:20001) manually as
a4c76882b99ef38e5f5ddd1d325de8cdca3cca02 (presubmit successful).

[commit-bot: I haz the power, 2016-03-03 15:29:17]

Description was changed from

==========
Move encoder thread to VideoSendStream.

Makes VideoCaptureInput easier to test and enables running more things
outside VideoCaptureInput on the encoder thread in the future
(initializing encoders and reconfiguring them, for instance).

BUG=webrtc:5410, webrtc:5494
R=stefan@webrtc.org

Committed:
https://chromium.googlesource.com/external/webrtc/+/a4c76882b99ef38e5f5ddd1d3...
==========

to

==========
Move encoder thread to VideoSendStream.

Makes VideoCaptureInput easier to test and enables running more things
outside VideoCaptureInput on the encoder thread in the future
(initializing encoders and reconfiguring them, for instance).

BUG=webrtc:5410, webrtc:5494
R=stefan@webrtc.org

Committed: https://crrev.com/a4c76882b99ef38e5f5ddd1d325de8cdca3cca02
Cr-Commit-Position: refs/heads/master@{#11860}
==========

[commit-bot: I haz the power, 2016-03-03 15:29:18]

Patchset 2 (id:??) landed as
https://crrev.com/a4c76882b99ef38e5f5ddd1d325de8cdca3cca02
Cr-Commit-Position: refs/heads/master@{#11860}