Re: stack usage in toast_insert_or_update()

On 1/30/07, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
> "Pavan Deolasee" <pavan(dot)deolasee(at)gmail(dot)com> writes:
> > The stack usage for toast_insert_or_update() may run into several KBs
> since
> > the MaxHeapAttributeNumber is set to a very large value of 1600. The
> usage
> > could anywhere between 28K to 48K depending on alignment and whether its
> a
> > 32-bit or a 64-bit machine.
>
> So? The routine is not re-entrant so I don't see that the stack space
> is a big problem. It's coded that way to avoid palloc/pfree cycles...

I always thought that it would be costlier to have a repeated stack
allocation/deallocation
of many KBs than dynamically allocating a small percentage of that. But I
might be wrong.
In fact, a small test I ran showed that mallloc/free is more costly. So may
be are
good.

Btw, I noticed that the toast_insert_or_update() is re-entrant.
toast_save_datum()
calls simple_heap_insert() which somewhere down the line calls
toast_insert_or_update() again. It looks a bit surprising, haven't look into
detail
though.

Thanks,
Pavan

--

EnterpriseDB http://www.enterprisedb.com