> From: Eric Hanchrow <eric.hanchrow@gmail.com>
> Date: Sat, 12 Dec 2015 22:37:08 +0000
>
> It's starting to look like "lui-adjust-undo-list" might be culpable; I think
> I've seen that function on the lisp stack each time.

Looks like another reincarnation of bug#21667.

Can you modify lui-adjust-undo-list so that GC is inhibited (by
binding gc-cons-threshold to most-positive-fixnum around the whole
function)?  It looks dangerous to me that this function messes with
the undo list inside mapconcat, which could cause GC, which could
decide to compact the current buffer, including shortening its undo
list, while lui-adjust-undo-list modifies it.