In copyBackingStore, when we reassign the newly copied backing store, we set the capacity to the size in bytes of the backing store. This makes us think we're way bigger than we actually are. Additionally, in visitChildren, we pass the size of the MapData to copyLater(), when we should be passing capacity * sizeof(Entry). I'll add some ASSERTs that should catch this sort of thing in the future.
Created attachment 211250 [details] Patch
Comment on attachment 211250 [details] Patch r=me Will the ASSERTs fire in our existing tests, or do we need a new test?
(In reply to comment #2) > (From update of attachment 211250 [details]) > r=me > > Will the ASSERTs fire in our existing tests, or do we need a new test? They fire in our existing tests.
Comment on attachment 211250 [details] Patch Clearing flags on attachment: 211250 Committed r155487: <http://trac.webkit.org/changeset/155487>
All reviewed patches have been landed. Closing bug.