Psql crashes with Segmentation fault on copy from

I am setting up a new machine and preparing some standard benchmark tests.
While trying to load some data using "copy from" psql is crashing.

OS Red Hat Enterprise Linux ES release 4 (Nahant Update 6)
Postgresql 8.2.7 installed from RPMs

I ran strace on psql and got:

read(5, "DAIRY QUEEN Q68\t5814\t1215"..., 4096) = 4096
rt_sigprocmask(SIG_BLOCK, NULL, [], 8) = 0
...... 36 other lines like the ones above/below.
rt_sigprocmask(SIG_BLOCK, NULL, [], 8) = 0
read(5, "TORE 2653\t5311\t93\t5080284\n2qLMfi"..., 4096) = 3796
rt_sigprocmask(SIG_BLOCK, NULL, [], 8) = 0
...... 36 other lines like the ones above/below.
rt_sigprocmask(SIG_BLOCK, NULL, [], 8) = 0
read(5, "", 4096) = 0
gettimeofday({1211915058, 706827}, NULL) = 0
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
+++ killed by SIGSEGV +++

On the server I increased the verbosity to debug levels:
DEBUG: 00000: checkpoint starting
LOCATION: CreateCheckPoint, xlog.c:5584
DEBUG: 00000: checkpoint complete; 0 transaction log file(s) added, 0
removed, 0 recycled
LOCATION: CreateCheckPoint, xlog.c:5712
LOG: 08P01: unexpected EOF on client connection
LOCATION: SocketBackend, postgres.c:307
DEBUG: 00000: proc_exit(0)
LOCATION: proc_exit, ipc.c:94
DEBUG: 00000: shmem_exit(0)
LOCATION: shmem_exit, ipc.c:125
LOG: 00000: disconnection: session time: 0:28:57.253 user=benchmark
database=benchmark host=[local]
LOCATION: log_disconnections, postgres.c:3840
DEBUG: 00000: exit(0)
LOCATION: proc_exit, ipc.c:112
DEBUG: 00000: server process (PID 3112) exited with exit code 0
LOCATION: LogChildExit, postmaster.c:2425

The same data file loads into another machine with Postgresql 8.2.4 and the
same version of RedHat..

Checked the disk with the badblocks command. No errors found.