On Thu, Apr 23, 2020 at 04:12:34PM -0400, James Coleman wrote:
>On Thu, Apr 23, 2020 at 8:38 AM Ranier Vilela <ranier.vf@gmail.com> wrote:
>>
>> Hi,
>>
>> Per Coverity.
>>
>> If has 0 full groups, "we don't need to do anything" and need goes to next.
>> Otherwise a integer division by zero, can raise.
>>
>> comments extracted trom explain.c:
>> /*
>> * Since we never have any prefix groups unless we've first sorted
>> * a full groups and transitioned modes (copying the tuples into a
>> * prefix group), we don't need to do anything if there were 0 full
>> * groups.
>> */
>
>This does look like a fairly obvious thinko on my part, and the patch
>looks correct to me.
>
>Tomas: agreed?
>
So how do we actually get the division by zero? It seems to me the fix
prevents a division by zero with 0 full groups and >0 prefix groups,
but can that actually happen?
But can that actually happen? Doesn't the comment quoted in the report
actually suggest otherwise? If this
(fullsortGroupInfo->groupCount == 0 &&
prefixsortGroupInfo->groupCount == 0)
evaluates to false, and
(fullsortGroupInfo->groupCount == 0)
this evaluates to true, then clearly there would have to be 0 full
groups and >0 prefix groups. But the comment says that can't happen,
unless I misunderstand what it's saying.
I've tried to trigger the issue, but without success ...
regards
--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services