From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 22398 invoked by alias); 25 Jun 2004 14:48:38 -0000 Mailing-List: contact binutils-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: binutils-owner@sources.redhat.com Received: (qmail 22389 invoked from network); 25 Jun 2004 14:48:37 -0000 Received: from unknown (HELO lucon.org) (24.6.43.109) by sourceware.org with SMTP; 25 Jun 2004 14:48:37 -0000 Received: by lucon.org (Postfix, from userid 1000) id DFD5E64D14; Fri, 25 Jun 2004 07:48:36 -0700 (PDT) Date: Fri, 25 Jun 2004 14:48:00 -0000 From: "H. J. Lu" To: Fabian Wenzel Cc: binutils@sources.redhat.com Subject: Re: libbfd crashes when linking in elf_i386_relocate_section Message-ID: <20040625144836.GA28920@lucon.org> References: <40DC29C3.9080603@tu-harburg.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <40DC29C3.9080603@tu-harburg.de> User-Agent: Mutt/1.4.1i X-SW-Source: 2004-06/txt/msg00447.txt.bz2 On Fri, Jun 25, 2004 at 03:33:55PM +0200, Fabian Wenzel wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Hi all! > > A month ago I already posted this error to gnu.utils.bug, but apparently > the list is not working currently (I wrote a follow-up today, but it did > not appear in the newsgroup). > > The problem is that the linker crashes when linking an own application with > > /usr/bin/ld: BFD 2.14.90.0.7 20031029 Debian GNU/Linux internal error, > aborting at ../../bfd/elf32-i386.c line 2247 in elf_i386_relocate_section > > I already tried binutils 2.15 which I built from the original tarball, > but the error remained. > > This problem suddently appeared; I did not make any changes to the code > which was building correctly for years. I did some debugging and found > out that the linker crashes as soon as two of three "critical" objects > are linked together. > > The abort() is called in the following lines: > > if (off >= (bfd_vma) -2) > ~ abort (); > > I added some tracing code and figured out that it is indeed caused by > one of the "critical" object files with section *UND*. > > If there is more that I can do in order to locate the problem, it would > be great if you could let me know. I am a little lost and don't know if > I am trying to figure things out in the right direction. Please open a bugzilla bug report and provide a small testcase. H.J.