Re: BUG #18483: Segmentation fault in tests modules

Masahiko Sawada <sawada.mshk@gmail.com>

From: Masahiko Sawada <sawada.mshk@gmail.com>
To: a.kozhemyakin@postgrespro.ru, pgsql-bugs@lists.postgresql.org
Date: 2024-06-10T00:33:50Z
Lists: pgsql-bugs
Hi,

Sorry for the late reply.

On Wed, May 29, 2024 at 1:41 AM PG Bug reporting form
<noreply@postgresql.org> wrote:
>
> The following bug has been logged on the website:
>
> Bug reference:      18483
> Logged by:          Alexander Kozhemyakin
> Email address:      a.kozhemyakin@postgrespro.ru
> PostgreSQL version: 17beta1
> Operating system:   ubuntu-22.04
> Description:
>
> second case
> echo "CREATE EXTENSION test_tidstore;
> SELECT do_set_block_offsets(1, ARRAY[0]::int2[]);" >
> src/test/modules/test_tidstore/sql/test_tidstore.sql
> make check -C src/test/modules/test_tidstore

Since the test script was modified so that it doesn't create tidstore
before setting TIDs, the server crashed with SEGV. I remember that we
used to check that the tidstore is already created before use, but we
removed that check because it would not be very helpful; if a
developer wiped out the create call then the test crashes pretty
obviously[1]. Do we want to revive that check as raising an error is
better than a SEGV?

Regards,

[1] https://www.postgresql.org/message-id/CANWCAZbJAxuw6HC%2BTc3p90mCZMKCaHknfrfzxMK5G0P6qp2PwQ%40mail.gmail.com

-- 
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com



Commits

  1. Make regress function make_tuple_indirect() able to handle plain attributes

  2. Fix segmentation fault in test_tidstore.

  3. Tighten test_predtest's input checks, and improve error messages.