Subject: misc/23374: Long-standing Intermittent "Undefined PLT symbol" problem
To: None <>
From: None <>
List: netbsd-bugs
Date: 11/05/2003 06:36:24
>Number:         23374
>Category:       misc
>Synopsis:       Long-standing Intermittent "Undefined PLT symbol" problem
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    misc-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Nov 05 06:37:00 UTC 2003
>Originator:     Torbjorn Granlund
>Release:        1.5 through 1.6.1
Swox AB
NetBSD 1.6.1 NetBSD 1.6.1 (GENERIC) #0: Wed Apr  9 03:02:04 UTC 2003 sparc

When doing lots of fork-execs, NetBSD intermittently reports
"Undefined PLT symbol".  Typical output:

/usr/pkg/bin/bash ../libtool --mode=compile gcc -DHAVE_CONFIG_H -I. -I/home/tege/prec/gmp-snap/mpz -I.. -D__GMP_WITHIN_GMP -I/home/tege/prec/gmp-snap    -O2 -mcpu=v7 -m32 -c -o init.lo `test -f '/home/tege/prec/gmp-snap/mpz/init.c' || echo '/home/tege/prec/gmp-snap/mpz/'`/home/tege/prec/gmp-snap/mpz/init.c
/usr/pkg/bin/bash: Undefined PLT symbol "close" (reloc type = 21, symnum = 180)
gmake[2]: *** [init.lo] Error 1
gmake[2]: Leaving directory `/var/tmp/gmp-obj/ss2/mpz'
gmake[1]: *** [all-recursive] Error 1
gmake[1]: Leaving directory `/var/tmp/gmp-obj/ss2'
gmake: *** [all] Error 2

I've run into this problem bulding gmp on powerpc, alpha, and sparc32
during the last years.  This bug has caused me to not run NetBSD on
almost any machine, but now I decided to try the latest release again.
The bug is still there.

How to run into this?  Get
and try to build it.  From googling a bit, it is clear that many people
have had this problem during the last 3 years with many other packages
than gmp.

Of course, intermittent problems can be a pain sometimes.  Fortunately,
this happens very often, at least on my machine.  It is actually not
possible to get a clean build of gmp without reissuing "make" at least
10 times.
Unfortunately, migrating away from netbsd is the only fix I know.
But if you actually react to this PR not by just closing it like
you did with PR 16388, 10455, and 9508, but by fixing the bug,
there is some hope for the future.  :-)