https://bugs.winehq.org/show_bug.cgi?id=57830
--- Comment #1 from Rafał Mużyło galtgendo@o2.pl --- ...after a few (dozen) iterations, I finally got that patch: it's correct due to awkward order of assignment/allocation: at the point it's allocated we don't know yet *if* there will be another item to assign, so we must allocate one more than it seems necessary, so we don't assign past the limit if we *do* get that item in the next loop. Bit tricky due to that order (and we might end up not needing any of the extra memory).