BUG #6392: leak memory while restore/load dump

The following bug has been logged on the website:

Bug reference: 6392
Logged by: Vic
Email address: vic(at)nix(dot)kh(dot)ua
PostgreSQL version: 9.0.6
Operating system: FreeBSD/Linux
Description:

While trying load database dump, postgresql (server process) eat up all
memory, including swap.

This problem start after update on 9.0.6 and same problem with 9.1.2.

PostgreSQL version 9.0.5 load same dump without problem.

In syslog on FreeBSD:
Jan 11 01:27:48 saturn kernel: swap_pager: out of swap space
Jan 11 01:27:49 saturn kernel: swap_pager_getswapspace(16): failed
Jan 11 01:27:49 saturn kernel: pid 2020 (postgres), uid 70, was killed: out
of swap space
Jan 11 01:27:49 saturn kernel: swap_pager: out of swap space
Jan 11 01:27:49 saturn kernel: swap_pager_getswapspace(16): failed
Jan 11 01:27:50 saturn postgres[2012]: [10-1] WARNING: terminating
connection because of crash of another server process
Jan 11 01:27:50 saturn postgres[2012]: [10-2] DETAIL: The postmaster has
commanded this server process to roll back the current transaction and exit,
because another server process exited abnormally and possibly corrupted
shared memory.
Jan 11 01:27:50 saturn postgres[2012]: [10-3] HINT: In a moment you should
be able to reconnect to the database and repeat your command.
Jan 11 01:27:50 saturn postgres[2036]: [10-1] WARNING: terminating
connection because of crash of another server process
Jan 11 01:27:50 saturn postgres[2036]: [10-2] DETAIL: The postmaster has
commanded this server process to roll back the current transaction and exit,
because another server process exited abnormally and possibly corrupted
shared memory.
Jan 11 01:27:50 saturn postgres[2036]: [10-3] HINT: In a moment you should
be able to reconnect to the database and repeat your command.

Notes: dump size in plain format about 650Mb, in c format about 125Mb;
database has plpython and postgis.