Don't fail on profile changes in GpuVideoDecoder.

Don't fail on profile changes in GpuVideoDecoder.

This restriction is artificial; or at least I can't make it fail
when the profile changes. The hardware decoders I've tried (Android,
Windows, Mac) don't seem to care if the profile actually changes.

In fact, MSE has always lied about the h264 profile (!), it always
specifies H264PROFILE_MAIN regardless of the actual profile. The
only parser sending this information correctly is the TS parser
used by Chromecast -- which then hits this check.

If the profile is actually unsupported we will still fail during
the check of SupportedProfiles.

BUG=640269
TEST=http://storage.googleapis.com/dalecurtis/mse_profile_change.html

Committed: https://crrev.com/3bd54ab36da9eb69239bfc95ee5e0e81ea3c42c0
Cr-Commit-Position: refs/heads/master@{#414164}

[DaleCurtis, 2016-08-23 18:19:14]

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

[DaleCurtis, 2016-08-23 18:19:20]

(testing now)

[DaleCurtis, 2016-08-23 19:08:27]

Description was changed from

==========
Don't fail on profile changes in GpuVideoDecoder.

This restriction seems artificial; or at least I can't make it fail
when the profile changes. The hardware decoders I've tried (Android,
Windows, Mac) don't seem to care if the profile actually changes.

If the profile is unsupported it will still fail during the check
of SupportedProfiles.

BUG=640269
TEST=manual
==========

to

==========
Don't fail on profile changes in GpuVideoDecoder.

This restriction seems artificial; or at least I can't make it fail
when the profile changes. The hardware decoders I've tried (Android,
Windows, Mac) don't seem to care if the profile actually changes.

If the profile is unsupported it will still fail during the check
of SupportedProfiles.

BUG=640269
TEST=http://storage.googleapis.com/dalecurtis/mse_profile_change.html
==========

[Tima Vaisburd, 2016-08-23 19:10:22]

On 2016/08/23 18:19:20, DaleCurtis wrote:
> (testing now)

What happens if adaptive playback is not supported?
https://cs.chromium.org/chromium/src/media/base/android/video_media_codec_dec...

[DaleCurtis, 2016-08-23 19:17:13]

Hah, I don't think it matters, because apparently MSE has been lying about the
profile all along :)

https://cs.chromium.org/chromium/src/media/formats/mp4/box_definitions.cc?l=671

[DaleCurtis, 2016-08-23 20:49:29]

Description was changed from

==========
Don't fail on profile changes in GpuVideoDecoder.

This restriction seems artificial; or at least I can't make it fail
when the profile changes. The hardware decoders I've tried (Android,
Windows, Mac) don't seem to care if the profile actually changes.

If the profile is unsupported it will still fail during the check
of SupportedProfiles.

BUG=640269
TEST=http://storage.googleapis.com/dalecurtis/mse_profile_change.html
==========

to

==========
Don't fail on profile changes in GpuVideoDecoder.

This restriction is artificial; or at least I can't make it fail
when the profile changes. The hardware decoders I've tried (Android,
Windows, Mac) don't seem to care if the profile actually changes.

In fact, MSE has always lied about the h264 profile (!), it always
specifies H264PROFILE_MAIN regardless of the actual profile. The
only parser sending this information correctly is the TS parser
used by Chromecast -- which then hits this check.

If the profile is actually unsupported we will still fail during
the check of SupportedProfiles.

BUG=640269
TEST=http://storage.googleapis.com/dalecurtis/mse_profile_change.html
==========

[DaleCurtis, 2016-08-23 20:51:31]

Pawel, can you check this out on a Chromebook? No other platforms seem to have
any real issues. Here's a link to the ES parser which does the right thing:

https://cs.chromium.org/chromium/src/media/formats/mp2t/es_parser_h264.cc?l=328

[DaleCurtis, 2016-08-24 00:19:15]

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

[DaleCurtis, 2016-08-24 00:19:16]

+wolenetz for MSE changes.

[wolenetz, 2016-08-24 00:48:34]

LGTM % some unit test (maybe mp4_stream_parser_unittests, as we chatted about).
Note that the VideoDecoderConfig::Matches() function will now see the distinct
profiles as indeed distinct, so SourceBufferStream will emit kConfigChanged when
there are two adjacent media segments that differ only in their avcC
profile_idc, making the world sane again :)

[wolenetz, 2016-08-24 00:52:03]

On 2016/08/24 00:48:34, wolenetz wrote:
> LGTM % some unit test (maybe mp4_stream_parser_unittests, as we chatted
about).
> Note that the VideoDecoderConfig::Matches() function will now see the distinct
> profiles as indeed distinct, so SourceBufferStream will emit kConfigChanged
when
> there are two adjacent media segments that differ only in their avcC
> profile_idc, making the world sane again :)

(And SourceBufferStream::UpdateVideoConfig() correctly does *not* error out on
just profile changes. At least that was sane previously :) )

[Pawel Osciak, 2016-08-24 06:55:34]

On 2016/08/23 20:51:31, DaleCurtis wrote:
> Pawel, can you check this out on a Chromebook? No other platforms seem to have
> any real issues. Here's a link to the ES parser which does the right thing:
> 
>
https://cs.chromium.org/chromium/src/media/formats/mp2t/es_parser_h264.cc?l=328

Yes, lgtm in general, as long as we keep checking for IsProfileSupported().
Profile information is in the stream, so the codecs should be able to handle the
change, as long as they support the particular profile.

[DaleCurtis, 2016-08-24 17:00:00]

The CQ bit was checked by dalecurtis@chromium.org

[DaleCurtis, 2016-08-24 17:00:01]

Thanks for the reviews; the cast folk would like this in branch today, so
landing as is and will add the test later this week.

[commit-bot: I haz the power, 2016-08-24 17:00:30]

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, 2016-08-24 20:13:25]

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

[commit-bot: I haz the power, 2016-08-24 20:13:29]

Try jobs failed on following builders:
  linux_android_rel_ng on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[DaleCurtis, 2016-08-24 20:30:25]

The CQ bit was checked by dalecurtis@chromium.org

[commit-bot: I haz the power, 2016-08-24 20:32:53]

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, 2016-08-24 22:04:36]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-08-24 22:06:43]

Description was changed from

==========
Don't fail on profile changes in GpuVideoDecoder.

This restriction is artificial; or at least I can't make it fail
when the profile changes. The hardware decoders I've tried (Android,
Windows, Mac) don't seem to care if the profile actually changes.

In fact, MSE has always lied about the h264 profile (!), it always
specifies H264PROFILE_MAIN regardless of the actual profile. The
only parser sending this information correctly is the TS parser
used by Chromecast -- which then hits this check.

If the profile is actually unsupported we will still fail during
the check of SupportedProfiles.

BUG=640269
TEST=http://storage.googleapis.com/dalecurtis/mse_profile_change.html
==========

to

==========
Don't fail on profile changes in GpuVideoDecoder.

This restriction is artificial; or at least I can't make it fail
when the profile changes. The hardware decoders I've tried (Android,
Windows, Mac) don't seem to care if the profile actually changes.

In fact, MSE has always lied about the h264 profile (!), it always
specifies H264PROFILE_MAIN regardless of the actual profile. The
only parser sending this information correctly is the TS parser
used by Chromecast -- which then hits this check.

If the profile is actually unsupported we will still fail during
the check of SupportedProfiles.

BUG=640269
TEST=http://storage.googleapis.com/dalecurtis/mse_profile_change.html

Committed: https://crrev.com/3bd54ab36da9eb69239bfc95ee5e0e81ea3c42c0
Cr-Commit-Position: refs/heads/master@{#414164}
==========

[commit-bot: I haz the power, 2016-08-24 22:06:43]

Patchset 3 (id:??) landed as
https://crrev.com/3bd54ab36da9eb69239bfc95ee5e0e81ea3c42c0
Cr-Commit-Position: refs/heads/master@{#414164}