Re: PG16.1 security breach?

From: Joe Conway <mail(at)joeconway(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Ron Johnson <ronljohnsonjr(at)gmail(dot)com>
Cc: "pgsql-general(at)lists(dot)postgresql(dot)org" <pgsql-general(at)lists(dot)postgresql(dot)org>
Subject: Re: PG16.1 security breach?
Date: 2024-06-13 11:59:20
Message-ID: 998b0cf7-d2f1-407a-965c-211cfc89ad47@joeconway.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On 6/12/24 18:56, Tom Lane wrote:
> Ron Johnson <ronljohnsonjr(at)gmail(dot)com> writes:
>> On Wed, Jun 12, 2024 at 4:36 PM David G. Johnston <
>> david(dot)g(dot)johnston(at)gmail(dot)com> wrote:
>>> I think my point is that a paragraph like the following may be a useful
>>> addition:
>>>
>>> If one wishes to remove the default privilege granted to public to execute
>>> all newly created procedures it is necessary to revoke that privilege for
>>> every superuser in the system
>
>> That seems... excessive.
>
> More to the point, it's wrong. Superusers have every privilege there
> is "ex officio"; we don't even bother to look at the catalog entries
> when considering a privilege check for a superuser. Revoking their
> privileges will accomplish nothing, and it does nothing about the
> actual source of the problem (the default grant to PUBLIC) either.
>
> What I'd do if I didn't like this policy is some variant of
>
> ALTER DEFAULT PRIVILEGES IN SCHEMA public
> REVOKE EXECUTE ON FUNCTIONS FROM PUBLIC;

In a past blog[1] I opined that this cleans up the default security
posture fairly completely:

8<----------------------
REVOKE CREATE ON SCHEMA public FROM PUBLIC;
REVOKE EXECUTE ON ALL ROUTINES IN SCHEMA public FROM PUBLIC;
ALTER DEFAULT PRIVILEGES IN SCHEMA public
REVOKE EXECUTE ON ROUTINES FROM PUBLIC;

-- And/or possibly, more drastic options:
-- REVOKE USAGE ON SCHEMA public FROM PUBLIC;
-- DROP SCHEMA public CASCADE;

REVOKE TEMPORARY ON DATABASE <your_db> FROM PUBLIC;
REVOKE USAGE ON LANGUAGE sql, plpgsql FROM PUBLIC;
8<----------------------

> Repeat for each schema that you think might be publicly readable
> (which is only public by default).

indeed

> BTW, in PG 15 and up, the public schema is not writable by
> default, which attacks basically the same problem from a different
> direction.

also a good point

[1]
https://www.crunchydata.com/blog/postgresql-defaults-and-impact-on-security-part-2

--
Joe Conway
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Rich Shepard 2024-06-13 12:37:17 Re: Definging columns for INSERT statements
Previous Message Daniel Gustafsson 2024-06-13 11:47:57 Re: Question about UNIX socket connections and SSL