Re: Odd 9.4, 9.3 buildfarm failure on s390x

I wrote:
> Here's a draft patch against HEAD for this.
> I looked for problem spots by (a) testing with the STRESS_SORT_INT_MIN
> option I added in nbtcompare.c, (b) grepping for "x = -x" type code,
> and (c) grepping for "return -x" type code. (b) and (c) found several
> places that (a) didn't, which does not give me a warm feeling about
> whether I have found quite everything.

I thought of another, uglier way to stress things: make wrappers around
memcmp, strcmp, and strncmp to force those to return INT_MIN/0/INT_MAX,
thereby modeling what we see is happening on Mark's machine.

This successfully exposed the bug I'd already found by grepping in
pg_rewind/filemap.c, as well as some astonishingly unportable code
in contrib/ltree.

The attached update incorporates Thomas' suggestion for the macro
name, as well as the ltree fix. For completeness, I also show the
very quick-hacky way I changed memcmp et al, but I don't propose
committing that.

I'm inclined to just go ahead and apply/backpatch this. It's certainly
possible that more bugs remain to be found, but I have no good ideas
about how to search for them, and in any case that wouldn't invalidate
the patch as it stands.

regards, tom lane