Bug 218549 – [GPU Process] Move the internal GPU rendering flags from WebPage to WebProcess

• Status: RESOLVED
• Resolution: FIXED
• Component: Canvas

[Said Abou-Hallawa]

Because not all the clients have access to a WebPage such as a worker based ImageBitmap.

[Said Abou-Hallawa]

Created attachment 413133 [details]
Patch

[Tim Horton]

Comment on attachment 413133 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=413133&action=review

> Source/WebKit/ChangeLog:8
> +        -- Setting UseGPUProcessForMedia is already in WebProcess.

This is wrong, not something to be emulated

> Source/WTF/Scripts/Preferences/WebPreferencesInternal.yaml:-697
> -  webcoreBinding: none

This is a layering violation.

> Source/WebKitLegacy/mac/WebView/WebPreferences.mm:3266
> +- (BOOL)useGPUProcessForCanvasRenderingEnabled

Clearly not.

[Said Abou-Hallawa]

Created attachment 413141 [details]
Patch

[Said Abou-Hallawa]

Comment on attachment 413133 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=413133&action=review

>> Source/WebKit/ChangeLog:8
>> +        -- Setting UseGPUProcessForMedia is already in WebProcess.
> 
> This is wrong, not something to be emulated

I tried to fix it but RemoteMediaPlayerManager::updatePreferences() uses settings.useGPUProcessForMediaEnabled(). So I kept it as is.

>> Source/WTF/Scripts/Preferences/WebPreferencesInternal.yaml:-697
>> -  webcoreBinding: none
> 
> This is a layering violation.

Change was reverted back.

>> Source/WebKitLegacy/mac/WebView/WebPreferences.mm:3266
>> +- (BOOL)useGPUProcessForCanvasRenderingEnabled
> 
> Clearly not.

Deleted.

[Simon Fraser (smfr)]

So should we fix the media flag?

[Said Abou-Hallawa]

(In reply to Simon Fraser (smfr) from comment #5)
> So should we fix the media flag?

I'll try to fix it.

[Said Abou-Hallawa]

Created attachment 413178 [details]
Patch

[Tim Horton]

I wonder if we need to fail harder if something tries to make two WebPages with different GPUP settings coexist in the same WebProcess?

[Tim Horton]

Comment on attachment 413178 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=413178&action=review

> Source/WTF/Scripts/Preferences/WebPreferencesInternal.yaml:719
> +  webcoreBinding: none

Nice!

[Sam Weinig]

If this is no longer per-page, WebPreferences is not the right API to use to surface this, as that is a per-page concept.

[Said Abou-Hallawa]

(In reply to Sam Weinig from comment #10)
> If this is no longer per-page, WebPreferences is not the right API to use to
> surface this, as that is a per-page concept.

I filed bug 218600 to track this issue.

[EWS]

Committed r269416: <https://trac.webkit.org/changeset/269416>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 413178 [details].

[Radar WebKit Bug Importer]

<rdar://problem/71065314>