The Hermit Hacker <scrappy(at)hub(dot)org> writes:
> How are we on this?
It's fixed.
I've also run the regress tests with bufmgr.c hacked up to discard
pages (with forcible overwriting) as soon as their refcount goes to
zero. That didn't disclose any similar bugs, although the coverage
of the tests leaves much to be desired :-(
regards, tom lane