From: Tom Lane <tgl@sss.pgh.pa.us>
To: Michael Paquier <michael@paquier.xyz>
cc: Daniel Gustafsson <daniel@yesql.se>,
        Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>,
        kuznetsovam@altlinux.org,
        PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>,
        egori@altlinux.org, nickel@altlinux.org
Subject: Re: pg_dump: Fix dangling pointer in EndCompressorZstd()
In-reply-to: <aAAvYUxZW7wibC66@paquier.xyz>
References: <efaee52b-9550-44ca-8633-ea86076b3283@altlinux.org>
 <CAExHW5uEAPLhaawqjZOU+yGjS1h8L_U0QGvrx+xA0Y3gF3hdSw@mail.gmail.com>
 <E85883B5-DAFA-40B9-9E0A-F2ED6A4144BA@yesql.se>
 <aAAvYUxZW7wibC66@paquier.xyz>
Comments: In-reply-to Michael Paquier <michael@paquier.xyz>
	message dated "Wed, 16 Apr 2025 15:29:53 -0700"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2085461.1744845172.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Wed, 16 Apr 2025 19:12:52 -0400
Message-ID: <2085462.1744845172@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/2085462.1744845172%40sss.pgh.pa.us>
Precedence: bulk

Michael Paquier <michael@paquier.xyz> writes:
> On Wed, Apr 16, 2025 at 04:19:02PM +0200, Daniel Gustafsson wrote:
>> Agreed, while it's perfectly safe today the end method should not make
>> assumptions about the use of the private_data pointer upon return and s=
hould
>> leave it set to NULL.

> Indeed.  I was just looking at applying what Alexander has sent
> because what EndCompressorZstd() not doing what the other methods do
> makes no sense.  Perhaps you are already on it, Daniel?

I think the actual reason for the difference is that the methods that
are taking care to zero the pointer do so because they test the
pointer themselves.  For instance in EndCompressorGzip, the test is
needed because perhaps no data was sent so the struct never got made.
It incidentally offers protection against a double call of that
function, but I don't think that was the intended reason.

I don't have any big objection to zeroing the pointer in
EndCompressorZstd, but I think the claim that it's precisely
analogous to the other EndCompressor methods is faulty,
because it has no similar test.

			regards, tom lane