I wrote:
> I haven't run it further to ground than that, but there's definitely
> something fishy here. Based on just these results one would be hard
> pressed to say if it's our bug or FreeBSD's, but your report that you
> don't see the failure with PG 11 makes it sound like our problem.
Ah, I have it: whoever wrote pg_GSS_read() failed to pay attention
to the fact that setting errno is a critical part of its API.
Sometimes it returns -1 while leaving errno in a state that causes
pqReadData to draw the wrong conclusions. In particular that can
happen when it reads an incomplete packet, and that's very timing
dependent, which is why this is so ticklish to reproduce.
I'll have a patch in a little while.
regards, tom lane