When constructing a permission prompt for WebRTC, a URI was supplied from the content process. This URI was untrusted, and could have been the URI of an origin that was previously granted permission; bypassing the prompt. This vulnerability affects F