Bug 217535 – [Media in GPU Process] Move AudioSessionRoutingArbitratorProxy to the GPU process

• Status: RESOLVED
• Resolution: FIXED
• Component: Media

[Peng Liu]

Currently AudioSessionRoutingArbitratorProxy is in the UI process.

[Radar WebKit Bug Importer]

<rdar://problem/70152548>

[Jer Noble]

Created attachment 424177 [details]
Patch

[Peng Liu]

Comment on attachment 424177 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=424177&action=review

> Source/WebKit/ChangeLog:13
> +        moved into WebCore to be used both there and in tihs new object.

s/tihs/this/g.

> Source/WebCore/platform/audio/mac/SharedRoutingArbitrator.h:53
> +    bool isInRoutingArbitrationForToken(Token&);

const Token&?

> Source/WebCore/platform/audio/mac/SharedRoutingArbitrator.mm:123
> +    ASSERT(isInRoutingArbitrationForToken(token));

Not a problem of this patch. Is it possible that this function is called when `m_setupArbitrationOngoing` is true?

> Source/WebKit/GPUProcess/mac/LocalAudioSessionRoutingArbitrator.cpp:47
> +

Nit. An empty line.

[Jer Noble]

Created attachment 424179 [details]
Patch

[Jer Noble]

Created attachment 424180 [details]
Patch

[Jer Noble]

(In reply to Peng Liu from comment #3)
> Comment on attachment 424177 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=424177&action=review
> 
> > Source/WebKit/ChangeLog:13
> > +        moved into WebCore to be used both there and in tihs new object.
> 
> s/tihs/this/g.

Whoops! Fixed.

> > Source/WebCore/platform/audio/mac/SharedRoutingArbitrator.h:53
> > +    bool isInRoutingArbitrationForToken(Token&);
> 
> const Token&?

Sure.

> > Source/WebCore/platform/audio/mac/SharedRoutingArbitrator.mm:123
> > +    ASSERT(isInRoutingArbitrationForToken(token));
> 
> Not a problem of this patch. Is it possible that this function is called
> when `m_setupArbitrationOngoing` is true?

The caller shouldn't end arbitration until the begin arbitration callback is complete.  We do handle that case, however.

> > Source/WebKit/GPUProcess/mac/LocalAudioSessionRoutingArbitrator.cpp:47
> > +
> 
> Nit. An empty line.

Removed.

[Jer Noble]

Created attachment 424181 [details]
Patch

[Eric Carlson]

Comment on attachment 424181 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=424181&action=review

r=me once the bots are happy

> Source/WebKit/GPUProcess/GPUConnectionToWebProcess.cpp:204
> +    , m_routingArbitrator(LocalAudioSessionRoutingArbitrator::create())

Is it possible to create this lazily?

> Source/WebKit/GPUProcess/mac/LocalAudioSessionRoutingArbitrator.cpp:57
> +    m_arbitrationUpdateTime = WallTime::now();

MonotonicTime?

> Source/WebKit/GPUProcess/mac/LocalAudioSessionRoutingArbitrator.h:54
> +    WallTime arbitrationUpdateTime() const { return m_arbitrationUpdateTime; }

Ditto.

[Jer Noble]

Created attachment 427679 [details]
Patch for landing

[Jer Noble]

Created attachment 427680 [details]
Patch for landing

[Jer Noble]

Created attachment 427697 [details]
Patch for landing

[EWS]

Committed r276993 (237316@main): <https://commits.webkit.org/237316@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 427697 [details].