From: Tom Lane <tgl@sss.pgh.pa.us>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
cc: Michael Paquier <michael@paquier.xyz>,
        Tomas Vondra <tomas.vondra@2ndquadrant.com>, exclusion@gmail.com,
        pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #16577: Segfault on altering a table located in a dropped
 tablespace
In-reply-to: <20201028134045.GA6709@alvherre.pgsql>
References: <20201028134045.GA6709@alvherre.pgsql>
Comments: In-reply-to Alvaro Herrera <alvherre@alvh.no-ip.org>
	message dated "Wed, 28 Oct 2020 10:40:45 -0300"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1251837.1603893564.1@sss.pgh.pa.us>
Date: Wed, 28 Oct 2020 09:59:24 -0400
Message-ID: <1251838.1603893564@sss.pgh.pa.us>
Precedence: bulk

Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> On 2020-Oct-28, Michael Paquier wrote:
>> Haven't thought of that approach, good idea!  That would not be
>> backpatchable but that would be a solution that does not require
>> creating files where we don't need them.  Did you begin to look at
>> that?

> I haven't started on this one yet, but I intend to do so shortly.

> Strictly speaking, we can still introduce a new category of pg_shdepend
> entries in back branches; it won't break anything that works today.

Yeah, as long as the patched version won't actively fail when those
pg_shdepend entries are missing, I don't think a backpatch is too
hazardous.  It might be worth checking that the extra entries don't
create huge problems if one does downgrade after some of them exist
--- but my feeling for how that mechanism works is that it'd Just
Work, and indeed provide the missing DROP protection even without
explicit action by the backend.

I would not be too excited about offering instructions for people
to manually add/remove the dependency entries.  The amount of
value added, versus the risks of bollixing things completely,
doesn't sound like a good tradeoff.

			regards, tom lane