Re: Unbounded %s in sscanf

From: Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>
To: Daniel Gustafsson <daniel(at)yesql(dot)se>
Cc: PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: Unbounded %s in sscanf
Date: 2021-06-28 16:45:10
Message-ID: 202106281645.eb555hja6lxs@alvherre.pgsql
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 2021-Jun-28, Daniel Gustafsson wrote:

> I happened to spot the below call in src/bin/pg_basebackup/streamutil.c which
> has an unbounded %s in the format.
>
> /* fetch xlog value and unit from the result */
> if (sscanf(PQgetvalue(res, 0, 0), "%d%s", &xlog_val, xlog_unit) != 2)
>
> There is no risk of overflow as the unit is defined to be at most 2 characters,
> but that's not explained (like how a similar %s is handled in pg_dump). The
> attached adds a small explanation in the comment to save readers time from
> following the bouncing ball to make sure.

Can you attack the system by crafting malicious output from the query?
I think the answer is still no, because the output comes from the query
SHOW wal_segment_size
which, if the attacker can control, the person running pg_basebackup has
way more serious problems.

But TBH it seems easy enough to limit to the output variable width.

--
Álvaro Herrera 39°49'30"S 73°17'W
"How amazing is that? I call it a night and come back to find that a bug has
been identified and patched while I sleep." (Robert Davidson)
http://archives.postgresql.org/pgsql-sql/2006-03/msg00378.php

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Andrew Dunstan 2021-06-28 17:02:37 Re: cleaning up PostgresNode.pm
Previous Message Tom Lane 2021-06-28 16:43:30 Re: Overflow hazard in pgbench