[webkit-reviews] review requested: [Bug 212695] [EME] Create CDMProxyFactory : [Attachment 401527] Patch

[bugzilla-daemon at webkit.org]

Xabier Rodríguez Calvar <calvaris at igalia.com> has asked  for review:
Bug 212695: [EME] Create CDMProxyFactory
https://bugs.webkit.org/show_bug.cgi?id=212695

Attachment 401527: Patch

https://bugs.webkit.org/attachment.cgi?id=401527&action=review




--- Comment #8 from Xabier Rodríguez Calvar <calvaris at igalia.com> ---
Created attachment 401527

  --> https://bugs.webkit.org/attachment.cgi?id=401527&action=review

Patch

(In reply to Charlie Turner from comment #7)
> I see this (some of the crashes are racey after apply this). The setProxy
> call has important timing preconditions which the ctor does not meet.
> 
> Regressions: Unexpected crashes (5)
>  
> imported/w3c/web-platform-tests/encrypted-media/clearkey-mp4-playback-
> temporary-waitingforkey.https.html [ Crash ]
>  
> imported/w3c/web-platform-tests/encrypted-media/clearkey-mp4-setmediakeys-
> again-after-resetting-src.https.html [ Crash ]
>  
> imported/w3c/web-platform-tests/encrypted-media/clearkey-mp4-syntax-
> mediakeys.https.html [ Crash ]
>  
> imported/w3c/web-platform-tests/encrypted-media/clearkey-mp4-syntax-
> mediakeysession.https.html [ Crash ]
>   media/encrypted-media/clearKey/clearKey-encrypted-cenc-event.html [ Crash ]

With the obsoleted patch these tests are flaky but not because of the key
stuff, only because we initialize GCrypt too soon. For that I created
LazilyInitialized and get that initialize when needed (I had the manual lazy
initalization version but writing a class for this seems possibly useful for
more people). I'm writing tests for it but I wanted to post it here to get
earlier reviews about it.