Re: _hash_addovflpage has a bug

On Mon, Jan 9, 2017 at 10:46 PM, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> Yeah, we can write code that way, but then it is better to rely just
> on retain_pin variable in the function and add an Assert for bucket
> page whenever we are retaining pin. How about doing something like
> attached patch?

Committed.

>> Not sure exactly how that
>> works out in terms of locking.
>
> We have to change the locking order as mentioned above by me. This
> change is already present in that patch, so maybe we add the check as
> suggested by you along with that patch. Now, another thing we could
> do is to extract those changes from WAL patch, but I am not sure if it
> is worth the effort.

I'm not sure at this point, either.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company