Re: ExecRTCheckPerms() and many prunable partitions (sqlsmith) - Mailing list pgsql-hackers

From Amit Langote
Subject Re: ExecRTCheckPerms() and many prunable partitions (sqlsmith)
Date
Msg-id CA+HiwqGAdVQh8PnMOPdjmF_9pcPsfSPXyNWVXZt=v6du1kSRTw@mail.gmail.com
Whole thread Raw
In response to Re: ExecRTCheckPerms() and many prunable partitions (sqlsmith)  (Justin Pryzby <pryzby@telsasoft.com>)
Responses Re: ExecRTCheckPerms() and many prunable partitions (sqlsmith)
List pgsql-hackers
On Mon, Feb 13, 2023 at 5:07 Justin Pryzby <pryzby@telsasoft.com> wrote:
On Tue, Nov 29, 2022 at 10:37:56PM +0900, Amit Langote wrote:
> 0002 contains changes that has to do with changing how we access
> checkAsUser in some foreign table planning/execution code sites.
> Thought it might be better to describe it separately too.

This was committed as 599b33b94:
    Stop accessing checkAsUser via RTE in some cases

That seems to add various elog()s which are hit frequently by sqlsmith:

postgres=# select from
(select transaction
from pg_prepared_xacts
right join pg_available_extensions
on false limit 0) where false;
ERROR:  permission info at index 2 (with relid=1262) does not match provided RTE (with relid=12081)

postgres=# select from (select confl_tablespace
from pg_stat_database_conflicts
where datname <> (select 'af')
limit 1) where false;
ERROR:  invalid perminfoindex 1 in RTE with relid 12271

Thanks for the report.  I’ll take a look once I’m back at a computer in a few days.
--
Thanks, Amit Langote
EDB: http://www.enterprisedb.com

pgsql-hackers by date:

Previous
From: Masahiko Sawada
Date:
Subject: Re: [PoC] Improve dead tuple storage for lazy vacuum
Next
From: Amit Kapila
Date:
Subject: Re: Rework LogicalOutputPluginWriterUpdateProgress