Re: PostgreSQL crashes with SIGSEGV

Andreas Seltenreich <andreas.seltenreich@credativ.de>

From: Andreas Seltenreich <andreas.seltenreich@credativ.de>
To: Michael Paquier <michael.paquier@gmail.com>
Cc: Bernd Helmle <mailings@oopsware.de>, Peter Geoghegan <pg@bowt.ie>, PostgreSQL mailing lists <pgsql-bugs@postgresql.org>
Date: 2017-12-14T18:47:08Z
Lists: pgsql-bugs, pgsql-hackers
Michael Paquier writes:

> Not sure if you can provide this information, but what does the plan
> of the query look like?

We did some more reducing work on the original query and data.  The
following testcase reproduces the double free reported by valgrind for
me when run against a vanilla REL9_6_STABLE build.

regards,
Andreas

--8<---------------cut here---------------start------------->8---
drop table if exists bug;
create table bug (n text, v text, b text, t text);
insert into bug
   select i%9, i%9, i%16 ,i%4096 from generate_series(1,100000) g(i);
analyze bug;

explain select * from (
select thecube.nv
   from ( select
            (n || ' ') || coalesce(v, '') as nv
           from bug
           group by ((n || ' ') || coalesce(v, '')) ,cube(b, t)
	  ) thecube
	 where nv = '8 8'
) sub limit 7000;
--8<---------------cut here---------------end--------------->8---

                                            QUERY PLAN
---------------------------------------------------------------------------------------------------
 Limit  (cost=13422.32..15747.28 rows=7000 width=32)
   ->  Subquery Scan on thecube  (cost=13422.32..27622.49 rows=42754 width=32)
         ->  GroupAggregate  (cost=13422.32..27194.95 rows=42754 width=38)
               Group Key: (((bug.n || ' '::text) || COALESCE(bug.v, ''::text))), bug.b, bug.t
               Group Key: (((bug.n || ' '::text) || COALESCE(bug.v, ''::text))), bug.b
               Group Key: (((bug.n || ' '::text) || COALESCE(bug.v, ''::text)))
               Sort Key: (((bug.n || ' '::text) || COALESCE(bug.v, ''::text))), bug.t
                 Group Key: (((bug.n || ' '::text) || COALESCE(bug.v, ''::text))), bug.t
               Filter: ((((bug.n || ' '::text) || COALESCE(bug.v, ''::text))) = '8 8'::text)
               ->  Sort  (cost=13422.32..13672.32 rows=100000 width=42)
                     Sort Key: (((bug.n || ' '::text) || COALESCE(bug.v, ''::text))), bug.b, bug.t
                     ->  Seq Scan on bug  (cost=0.00..2041.00 rows=100000 width=42)



Commits

  1. Fix actual and potential double-frees around tuplesort usage.

  2. Allow the built-in ordered-set aggregates to share transition state.

  3. Avoid integer overflow while sifting-up a heap in tuplesort.c.

  4. Allow avoiding tuple copy within tuplesort_gettupleslot().

  5. Remove should_free arguments to tuplesort routines.

  6. Fix use-after-free around DISTINCT transition function calls.

  7. Reuse abbreviated keys in ordered [set] aggregates.