IMPORTANT: two new PostgreSQL security problems found

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: pgsql-announce(at)postgresql(dot)org
Subject: IMPORTANT: two new PostgreSQL security problems found
Date: 2005-05-02 20:06:38
Message-ID: 2369.1115064398@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin pgsql-announce pgsql-general

Two serious security errors have been found in PostgreSQL 7.3 and newer
releases. These errors at least allow an unprivileged database user to
crash the backend process, and may make it possible for an unprivileged
user to gain the privileges of a database superuser.

We are currently preparing new releases that will correct these problems
in freshly initdb'd installations. However, because these problems are
really incorrect system catalog entries, updating to a new release will
NOT by itself solve the problems in an existing installation. Instead,
it is necessary for the database administrator to fix the catalog entries
manually, as described below. We are releasing this advisory to encourage
administrators of PostgreSQL installations to perform these fixes as soon
as possible.

Character conversion vulnerability
----------------------------------

The more severe of the two errors is that the functions that support
client-to-server character set conversion can be called from SQL commands
by unprivileged users, but these functions are not designed to be safe
against malicious choices of argument values. This problem exists in
PostgreSQL 7.3.* through 8.0.*. The recommended fix is to disable public
EXECUTE access for these functions. This does not affect normal usage of
the functions for character set conversion, but it will prevent misuse.

To accomplish this change, execute the following SQL command as a
superuser:

UPDATE pg_proc SET proacl = '{=}'
WHERE pronamespace = 11 AND pronargs = 5
AND proargtypes[2] = 'cstring'::regtype;

In 7.3.* through 8.0.*, this should report having updated 90 rows.
7.4 and later will report a "WARNING: defaulting grantor to user ID 1"
which can be ignored.

The above command must be carried out in *each* database of an
installation, including template1, and ideally including template0 as
well. If you do not fix the template databases then any subsequently
created databases will contain the same vulnerability. template1 can
be fixed in the same way as any other database, but fixing template0
requires additional steps. First, from any database issue

UPDATE pg_database SET datallowconn = true WHERE datname = 'template0';

Next connect to template0 and perform the pg_proc update. Finally, do

-- re-freeze template0:
VACUUM FREEZE;
-- and protect it against future alterations:
UPDATE pg_database SET datallowconn = false WHERE datname = 'template0';

tsearch2 vulnerability
----------------------

The other error is that the contrib/tsearch2 module misdeclares several
functions as returning type "internal" when they do not have any
"internal" argument. This breaks the type safety of "internal" by
allowing users to construct SQL commands that invoke other functions
accepting "internal" arguments. The consequences of this have not been
investigated in detail, but it is certainly at least possible to crash
the backend.

This error affects PostgreSQL 7.4 and later, but only if you have
installed the contrib/tsearch2 module. The recommended fix is to
change the misdeclared functions so that they accept an "internal"
argument and therefore cannot be called directly from SQL commands.
To do this, execute the following command as a superuser:

UPDATE pg_proc SET proargtypes[0] = 'internal'::regtype
WHERE oid IN (
'dex_init(text)'::regprocedure,
'snb_en_init(text)'::regprocedure,
'snb_ru_init(text)'::regprocedure,
'spell_init(text)'::regprocedure,
'syn_init(text)'::regprocedure
);

This should report 5 rows updated. (If it fails with a message
like "function "dex_init(text)" does not exist", then either tsearch2
is not installed in this database, or you already did the update.)

You will need to do this in *each* database in which you have installed
tsearch2, including template1. You need not worry about template0,
however, since it will certainly not contain tsearch2.

If you frequently install tsearch2 in new databases, you will also
want to modify the tsearch.sql script to declare these functions as
taking type internal in the first place. (The script fix will be part
of the upcoming releases, so you may be able to wait for those.)

On behalf of the PostgreSQL core committee, I'd like to apologize for
any problems that may arise from these errors.

regards, tom lane

Responses

Browse pgsql-admin by date

  From Date Subject
Next Message Keith Jones 2005-05-02 21:10:29 Keith Jones is out of the office.
Previous Message Tom Lane 2005-05-02 14:52:47 Re: A nightmare

Browse pgsql-announce by date

  From Date Subject
Next Message Thomas F.O'Connell 2005-05-02 22:09:13 Re: [ANNOUNCE] IMPORTANT: two new PostgreSQL security problems found
Previous Message Enrico Weigelt 2005-05-02 03:55:53 Re: pgtop, display PostgreSQL processes in `top' style

Browse pgsql-general by date

  From Date Subject
Next Message Tom Lane 2005-05-02 20:25:02 Re: 7.3.9 Install Question - init.d/postgresql error?
Previous Message Dianne Chen 2005-05-02 19:22:31 Re: 7.3.9 Install Question - init.d/postgresql error?