Handle rotation correctly in WebRTC GPUProcess rendering pipeline

• Component: WebRTC
• Status: RESOLVED FIXED

[youenn fablet]

Handle rotation correctly in WebRTC GPUProcess rendering pipeline

[youenn fablet]

<rdar://76915991>

[youenn fablet]

Created attachment 426802 [details]
Patch

[Eric Carlson]

Comment on attachment 426802 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=426802&action=review

> Source/WebKit/GPUProcess/webrtc/RemoteSampleBufferDisplayLayer.cpp:81
> -    m_sampleBufferDisplayLayer->updateAffineTransform(transform);
> +    m_sampleBufferDisplayLayer->updateRootLayerAffineTransform(transform);

Why don't we have to do this in the web process? 

What happens in the web process if `updateAffineTransform`to rotates the root layer?

[youenn fablet]

(In reply to Eric Carlson from comment #3)
> Comment on attachment 426802 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=426802&action=review
> 
> > Source/WebKit/GPUProcess/webrtc/RemoteSampleBufferDisplayLayer.cpp:81
> > -    m_sampleBufferDisplayLayer->updateAffineTransform(transform);
> > +    m_sampleBufferDisplayLayer->updateRootLayerAffineTransform(transform);
> 
> Why don't we have to do this in the web process? 
> 
> What happens in the web process if `updateAffineTransform`to rotates the
> root layer?

I am not quite sure why doing it in the root layer and not the sample layer fixes the rotation issue in GPUProcess.
If I use the root layer in WebProcess, rotation is fine but the internal sample buffer layer is being cropped.

I asked around and it would be good to have additional analysis in that area.
In the meantime, I am landing this patch now since it fixes the issue.

[youenn fablet]

iOS-wk2 failure is unrelated.

[EWS]

Committed r276497 (236956@main): <https://commits.webkit.org/236956@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 426802 [details].