Re: init_htab causes SIGFPE (or worse) due to miscalculation for large nbuckets

Jeff Davis <pgsql@j-davis.com> writes:
> On Mon, 2012-12-10 at 20:27 -0500, Tom Lane wrote:
>> That does seem like a good idea.  We need one for an int-sized result
>> too, to fix the original problem in init_htab.  So I propose these
>> functions:

> Looks good to me. One other corner case in the version of the patch I
> was working on was that nbuckets is compared to num_partitions, which is
> a long. We can assert that it is less than or equal to INT_MAX in
> hash_create.

> Aside from that, I'll drop my version of the patch, which doesn't have
> any useful differences from yours.

I hadn't gone any further than to code and test the functions I listed.
If you are working on a complete patch, please press on.

            regards, tom lane