Re: Avoid a possible out-of-bounds access (src/backend/optimizer/util/relnode.c)

From: Etsuro Fujita <etsuro(dot)fujita(at)gmail(dot)com>
To: Ranier Vilela <ranier(dot)vf(at)gmail(dot)com>
Cc: Pg Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Avoid a possible out-of-bounds access (src/backend/optimizer/util/relnode.c)
Date: 2023-09-24 04:50:39
Message-ID: CAPmGK15uitYHTmH=Z7pwvrHKW+08YWZoUYbGy_Ecym54DdLJ6w@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

On Sat, Sep 23, 2023 at 9:59 PM Ranier Vilela <ranier(dot)vf(at)gmail(dot)com> wrote:
> Per Coverity.
> CID 1518088 (#2 of 2): Improper use of negative value (NEGATIVE_RETURNS)
>
> The function bms_singleton_member can returns a negative number.
>
> /*
> * Get a child rel for rel2 with the relids. See above comments.
> */
> if (rel2_is_simple)
> {
> int varno = bms_singleton_member(child_relids2);
>
> child_rel2 = find_base_rel(root, varno);
> }
>
> It turns out that in the get_matching_part_pairs function (joinrels.c), the return of bms_singleton_member is passed to the find_base_rel function, which cannot receive a negative value.
>
> find_base_rel is protected by an Assertion, which effectively indicates that the error does not occur in tests and in DEBUG mode.
>
> But this does not change the fact that bms_singleton_member can return a negative value, which may occur on some production servers.
>
> Fix by changing the Assertion into a real test, to protect the simple_rel_array array.

Thanks for the report and patch! I will review the patch.

Best regards,
Etsuro Fujita

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Kuwamura Masaki 2023-09-24 08:22:23 Re: bug fix and documentation improvement about vacuumdb
Previous Message Michael Paquier 2023-09-24 01:42:49 Re: [HACKERS] Should logtape.c blocks be of type long?