Re: memory-related bugs

Bruce Momjian <bruce@momjian.us> writes:
> Did we conclude any of these were useful?
>     http://archives.postgresql.org/pgsql-hackers/2011-03/msg00856.php
> I know there were concerns about some of them in the thread.

Hmm, I guess this slipped through the cracks.  I thought that avoiding
memcpy(x, x, n) was unnecessary, and I had doubts about the style of
some of the other changes, but I think we do need to avoid accessing
past the defined end of a data structure.  We've seen cases in the past
where one day that structure is right up against the end of memory and
you get a SIGSEGV; there's no good reason to believe it cannot happen
in these places.
        regards, tom lane