Set audible status to false if audio is stopped after prefetch.

Set audible status to false if audio is stopped after prefetch.

We discovered an audio playing site that often triggers
the starvation condition during the playback. This introduces
a race between Pause , starvation detecton and the MediaDecoderCallback.
If starvation detection happens between Pause and the callback,
we will do Prefetch and then stop the flow of packets,
and won't update the audible status properly.

This fix sets the audible status to false in Pause()
and this avoids the issue.

BUG=474761
Committed: https://crrev.com/9bf805c15752621d68d562f2acfa3e71f98f5593
Cr-Commit-Position: refs/heads/master@{#327336}

[timav, 2015-04-07 22:18:08]

timav@google.com changed reviewers:
+ qinmin@chromium.org, timav@google.com

[timav, 2015-04-07 22:18:08]

Please review.

[qinmin, 2015-04-14 02:55:11]

https://codereview.chromium.org/1067153002/diff/1/media/base/android/media_so...
File media/base/android/media_source_player.cc (right):

https://codereview.chromium.org/1067153002/diff/1/media/base/android/media_so...
media/base/android/media_source_player.cc:501: if (is_audio)
how about move this to line 485 ? also, what about calling SetAudible(false) in
Pause(), we don't need very accurate timings on the audio notification.

[timav, 2015-04-28 02:17:55]

https://codereview.chromium.org/1067153002/diff/1/media/base/android/media_so...
File media/base/android/media_source_player.cc (right):

https://codereview.chromium.org/1067153002/diff/1/media/base/android/media_so...
media/base/android/media_source_player.cc:501: if (is_audio)
On 2015/04/14 02:55:10, qinmin wrote:
> how about move this to line 485 ? also, what about calling SetAudible(false)
in
> Pause(), we don't need very accurate timings on the audio notification.

I cannot do that because of false starvations which send
PREFETCH_REQUEST_EVENT_PENDING. EVENT_PENDING is immediately creared and we get
flickering. So far I was unable to stop these false starvations.

Calling SetAudible(false) in Pause seems to solve the problem, shall we do that
instead?

[qinmin, 2015-04-28 16:12:24]

lgtm

[timav, 2015-04-28 17:42:44]

The CQ bit was checked by timav@google.com

[commit-bot: I haz the power, 2015-04-28 17:42:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1067153002/20001

[commit-bot: I haz the power, 2015-04-28 18:36:10]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-04-28 18:36:59]

Patchset 2 (id:??) landed as
https://crrev.com/9bf805c15752621d68d562f2acfa3e71f98f5593
Cr-Commit-Position: refs/heads/master@{#327336}