[Media In GPU Process][MSE] Add infrastructure needed to run MediaPlayerPrivateMediaSourceAVFObjC in the GPU process

• Component: Media
• Status: RESOLVED FIXED

[Peng Liu]

We need to add MediaSourcePrivateRemote/RemoteMediaSourceProxy and SourceBufferPrivateRemote/RemoteSourceBufferProxy and setup the IPC messages.

[Peng Liu]

Created attachment 414238 [details]
WIP patch

[Peng Liu]

I need to rebase this patch.

[Peng Liu]

Created attachment 414246 [details]
Rebased WIP patch

[Peng Liu]

Created attachment 414253 [details]
WIP patch: enable unified build

[Peng Liu]

Created attachment 414255 [details]
WIP patch: fix WPE build failures

[Peng Liu]

Created attachment 414258 [details]
WIP patch: fix WPE build failures again

[Peng Liu]

Created attachment 414265 [details]
WIP patch: fix build failures

[Peng Liu]

Created attachment 414276 [details]
WIP patch: fix build failures on WPE and GTK ports

[Peng Liu]

Created attachment 414277 [details]
WIP patch: fix build failures on WPE and GTK ports (again)

[Peng Liu]

Created attachment 414280 [details]
WIP patch: fix build failures on WPE and GTK ports

[Peng Liu]

Created attachment 414285 [details]
WIP patch: fix build failures on WPE and GTK ports (update WebKit_MESSAGES_IN_FILES in CMakeLists.txt)

[Peng Liu]

Created attachment 414286 [details]
Patch

[Eric Carlson]

Comment on attachment 414286 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=414286&action=review

> Source/WebCore/platform/graphics/MediaPlayer.cpp:-480
> -    m_contentType = contentType;

Is this really not needed if we end up in `nextBestMediaEngine`?

> Source/WebKit/GPUProcess/media/RemoteMediaSourceProxy.cpp:61
> +    return { };

You could call `notImplemented()` from all of these stub methods.

> Source/WebKit/GPUProcess/media/RemoteSourceBufferProxy.cpp:62
> +}

Ditto

> Source/WebKit/WebProcess/GPU/media/MediaPlayerPrivateRemote.cpp:637
> +    if (m_remoteEngineIdentifier == MediaPlayerEnums::MediaEngineIdentifier::AVFoundationMSE) {
> +        auto identifier = RemoteMediaSourceIdentifier::generate();
> +        connection().send(Messages::RemoteMediaPlayerProxy::LoadMediaSource(url, identifier), m_id);
> +        m_mediaSourcePrivate = MediaSourcePrivateRemote::create(m_manager.gpuProcessConnection(), identifier, m_manager.typeCache(m_remoteEngineIdentifier), *this, client);
> +    } else {

You could use an early return here instead of the `else`

> Source/WebKit/WebProcess/GPU/media/MediaSourcePrivateRemote.cpp:107
> +}

notImplemented()

> Source/WebKit/WebProcess/GPU/media/SourceBufferPrivateRemote.cpp:85
> +}

Ditto

[Peng Liu]

Created attachment 414304 [details]
Patch for landing

[Peng Liu]

Comment on attachment 414286 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=414286&action=review

>> Source/WebCore/platform/graphics/MediaPlayer.cpp:-480
>> -    m_contentType = contentType;
> 
> Is this really not needed if we end up in `nextBestMediaEngine`?

Hmm, we may need it in that case. To be safe, I added the parameter back and changed MediaPlayerPrivate::load() accordingly.

>> Source/WebKit/GPUProcess/media/RemoteMediaSourceProxy.cpp:61
>> +    return { };
> 
> You could call `notImplemented()` from all of these stub methods.

Fixed.

>> Source/WebKit/GPUProcess/media/RemoteSourceBufferProxy.cpp:62
>> +}
> 
> Ditto

Fixed.

>> Source/WebKit/WebProcess/GPU/media/MediaPlayerPrivateRemote.cpp:637
>> +    } else {
> 
> You could use an early return here instead of the `else`

Right! Fixed.

>> Source/WebKit/WebProcess/GPU/media/MediaSourcePrivateRemote.cpp:107
>> +}
> 
> notImplemented()

Fixed.

>> Source/WebKit/WebProcess/GPU/media/SourceBufferPrivateRemote.cpp:85
>> +}
> 
> Ditto

Fixed.

[EWS]

Committed r269907: <https://trac.webkit.org/changeset/269907>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 414304 [details].

[Radar WebKit Bug Importer]

<rdar://problem/71489527>