On Sun, Apr 11, 2021 at 10:55 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Alternatively, we could do what you suggested and redefine things
> so that one is only allowed to set the all-visible bit while holding
> superexclusive lock; which again would allow an enormous simplification
> in heap_update and cohorts.
Great detective work.
I would rather not go back to requiring a superexclusive lock in
vacuumlazy.c (outside of pruning), actually -- I was only pointing out
that that had changed, and was likely to be relevant. It wasn't a real
proposal.
I think that it would be hard to justify requiring a super-exclusive
lock just to call PageSetAllVisible(). PD_ALL_VISIBLE is fundamentally
redundant information, so somehow it feels like the wrong design.
> Either way, it's hard to argue that
> heap_update hasn't crossed the complexity threshold where it's
> impossible to maintain safely. We need to simplify it.
It is way too complicated. I don't think that I quite understand your
first proposal right now, so I'll need to go think about it.
--
Peter Geoghegan
