BUG #3648: Server crashes when trying to create a table

The following bug has been logged online:

Bug reference:      3648
Logged by:          Anton
Email address:      anton@orkney.co.jp
PostgreSQL version: 8.2.5 and 8.1.9
Operating system:   CentOS 5 and Xubuntu 7.04
Description:        Server crashes when trying to create a table
Details:

I have a plpgsql function which creates a table with following line:

CREATE TABLE vertices_tmp(id serial);

It's working fine with PostgreSQL 8.2.4 but crashes with segfault at
versions 8.1.9 and 8.2.5.

Here is a piece of log file:

2007-10-03 13:15:13 JST NOTICE:  CREATE TABLE will create implicit sequence
"vertices_tmp_id_seq" for serial column "vertices_tmp.id"
2007-10-03 13:15:13 JST CONTEXT:  SQL statement "CREATE TABLE "vertices_tmp"
( id serial )"
    PL/pgSQL function "assign_vertex_id" line 19 at SQL statement
2007-10-03 13:15:13 JST LOG:  server process (PID 10424) was terminated by
signal 11
2007-10-03 13:15:13 JST LOG:  terminating any other active server processes
2007-10-03 13:15:13 JST LOG:  all server processes terminated;
reinitializing
2007-10-03 13:15:13 JST LOG:  database system was interrupted at 2007-10-03
13:14:00 JST
2007-10-03 13:15:13 JST LOG:  checkpoint record is at 0/A27F48
2007-10-03 13:15:13 JST LOG:  redo record is at 0/A27F48; undo record is at
0/0; shutdown FALSE
2007-10-03 13:15:13 JST LOG:  next transaction ID: 0/877; next OID: 49152
2007-10-03 13:15:13 JST LOG:  next MultiXactId: 1; next MultiXactOffset: 0
2007-10-03 13:15:13 JST LOG:  database system was not properly shut down;
automatic recovery in progress
2007-10-03 13:15:13 JST FATAL:  the database system is starting up
2007-10-03 13:15:13 JST LOG:  redo starts at 0/A27F90
2007-10-03 13:15:13 JST LOG:  invalid magic number 0000 in log file 0,
segment 0, offset 10960896
2007-10-03 13:15:13 JST LOG:  redo done at 0/A73728
2007-10-03 13:15:13 JST LOG:  database system is ready