ResourceLoader::willCacheResponse() needs to null-check m_frame->settings() We've seen crash reports at Apple (<rdar://problem/7249553>) where this method is called on a ResourceLoader, that belongs to a Frame, that no longer has a Page. Nominally, we think that any Frame without a Page shouldn't have loads occuring, but there is at least one code path where that is not the case, and it's proving elusive to figure out that code path by code inspection. Plenty of sites in the code null check frame->settings(), and we need to do so here to prevent this crash.
Created attachment 58947 [details] Patch v1 (null check and ASSERT)
http://trac.webkit.org/changeset/61296