Date: Sun, 13 Oct 2019 22:38:48 +0200
From: Tomas Vondra <tomas.vondra@2ndquadrant.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Andres Freund <andres@anarazel.de>, buschmann@nidsa.net,
	pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #16045: vacuum_db crash and illegal memory alloc after
 pg_upgrade from PG11 to PG12
Message-ID: <20191013203848.2eksimmpux3546wr@development>
References: <20191009232836.jg77wccqwy5fgs6s@development>
 <20265.1570664514@sss.pgh.pa.us>
 <20191010014813.3b2btmkakhrskwgd@alap3.anarazel.de>
 <20191010134303.bkwjzchjb2w66iro@development>
 <30706.1570717152@sss.pgh.pa.us>
 <20191010163350.2rtcjtje6tkgilqi@development>
 <1542.1570738460@sss.pgh.pa.us>
 <20191010204022.6hiiubm2cueu3ghk@development>
 <20191013001032.lqrn4kqaovhrf77l@development>
 <17717.1570991208@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Disposition: inline
In-Reply-To: <17717.1570991208@sss.pgh.pa.us>
User-Agent: NeoMutt/20180716
Precedence: bulk

On Sun, Oct 13, 2019 at 02:26:48PM -0400, Tom Lane wrote:
>Tomas Vondra <tomas.vondra@2ndquadrant.com> writes:
>> here is an updated patch, with the recursive CTE. I've done a fair
>> amount of testing on it on older versions (up to 9.4), and it seems to
>> work just fine.
>
>Might be a good idea to exclude attisdropped columns in the part of the
>recursive query that's looking for sql_identifier columns of composite
>types.  I'm not sure if composites can have dropped columns today,
>but even if they can't it seems like a wise bit of future-proofing.
>(We'll no doubt have occasion to use this logic again...)
>

Hmm? How could that be safe? Let's say we have a composite type with a
sql_identifier column, it's used in a table with data, and we drop the
column. We need the pg_type information to parse the existing, so how
could we skip attisdropped columns?

>Looks good other than that nit.
>
>> BTW the query (including the RELKIND_COMPSITE_TYPE) was essentially just
>> a lightly-massaged copy of old_9_6_check_for_unknown_data_type_usage, so
>> that seems wrong too.
>
>Yeah, we should back-port this logic into that check too, IMO.
>

You mean the recursive CTE, removal of RELKIND_COMPOSITE_TYPE or the
proposed change w.r.t. dropped columns?


regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services