Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> So, the only callers of both has already acquired appropiate locks at
> the relation level -- nobody is going to be modifying the blocks while
> they proceed. So why bother locking the pages at all? Is there a
> reason or is this an historical accident?
No, because operations such as checkpointing and bgwriter will feel free
to write out pages that aren't exclusive-locked; they don't try to get
a lock at the table level. Failing to lock the buffer would risk
allowing an invalid page state to be written to disk --- which, if we
then crashed before writing the WAL record for the vacuum operation,
would represent unrecoverable corruption.
regards, tom lane