Created attachment 264523 [details]
Patch

Comment on attachment 264523 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=264523&action=review

> Source/WebCore/page/ResourceUsageOverlay.cpp:60
> +    FrameView& frameView = *m_page.mainFrame().view();

What guarantees the view is non-null?

(In reply to comment #2)
> Comment on attachment 264523 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=264523&action=review
> 
> > Source/WebCore/page/ResourceUsageOverlay.cpp:60
> > +    FrameView& frameView = *m_page.mainFrame().view();
> 
> What guarantees the view is non-null?

Through the magic of the initial empty document load, which happens by way of WebKit calling into Frame::init() on the MainFrame during WebPage bringup, we end up constructing a FrameView when transitioning to committed state.

This all happens before the ResourceUsageOverlay is constructed (when the setting is flipped to true, either during WebPage::updatePreferences(), or at any time later through WebCore::Settings API.)

Of course, none of this is remotely obvious from looking at the pointer dereference in context, so I'll just put in a null check. :)

Committed r191861: <http://trac.webkit.org/changeset/191861>