Re: pgsql: Add checks to TRUNCATE, CLUSTER, and REINDEX to prevent

Bruce Momjian <bruce@momjian.us> writes:
> You want a TODO for 3883?

Yeah, I would like to have some positive detection for deadlocks
involving LockBufferForCleanup().  What is unclear is how the heck
to do that without adding unacceptable overhead to buffer pin/unpin.

            regards, tom lane