From: Andrew Cagney <cagney@gnu.org>
To: Alan Modra <amodra@bigpond.net.au>
Cc: binutils@sources.redhat.com
Subject: Re: BFD: /home/cagney/tmp/a.out(.rela.plt): relocation 0 has invalid symbol index 1
Date: Wed, 26 Jan 2005 20:33:00 -0000 [thread overview]
Message-ID: <41F7FE5F.3060703@gnu.org> (raw)
In-Reply-To: <20050125103011.GC17250@bubble.modra.org>
Alan Modra wrote:
> On Tue, Jan 25, 2005 at 08:23:59PM +1030, Alan Modra wrote:
>
>>On Mon, Jan 24, 2005 at 04:51:12PM -0500, Andrew Cagney wrote:
>>
>>>$ nm --synthetic ...
>>>BFD: /home/cagney/tmp/a.out(.rela.plt): relocation 0 has invalid symbol
>>>index 1
>>>10010900 ? *ABS*@plt
>>>10010918 B bar
>>
>>It's a bug. _bfd_elf_get_synthetic_symtab is trying to slurp dynamic
>>relocs before the dynamic symbol table has been read.
>>
>>"nm -D --synthetic" works. It looks like "nm --synthetic" is supposed
>>to work too. Oh well, fixing.
>
>
> As well as fixing the above, I've restricted the relocs returned by
> _bfd_elf_canonicalize_dynamic_reloc to those for loadable sections.
> The idea is to exclude debug relocs, and any other weird stuff that
> generally isn't of interest for "objdump -dR".
Works for me.
> I see gdb makes
> calls to bfd_canonicalize_dynamic_reloc too, but I think gdb/dbxread.c
> and gdb/solib-frv.c aren't interested in debug relocs either.
Agreed.
thanks!
Andrew
>
> bfd/
> * elf.c (_bfd_elf_get_dynamic_reloc_upper_bound): Only include
> loadable reloc sections.
> (_bfd_elf_canonicalize_dynamic_reloc): Likewise.
> (_bfd_elf_get_synthetic_symtab): Return 0 if no dynamic syms.
> binutils/
> * nm.c (display_rel_file): Read dynamic syms before calling
> bfd_get_synthetic_symtab.
>
> Index: bfd/elf.c
> ===================================================================
> RCS file: /cvs/src/src/bfd/elf.c,v
> retrieving revision 1.260
> diff -u -p -r1.260 elf.c
> --- bfd/elf.c 7 Jan 2005 09:52:00 -0000 1.260
> +++ bfd/elf.c 25 Jan 2005 10:07:51 -0000
> @@ -5977,10 +5977,10 @@ _bfd_elf_canonicalize_dynamic_symtab (bf
> return symcount;
> }
>
> -/* Return the size required for the dynamic reloc entries. Any
> - section that was actually installed in the BFD, and has type
> - SHT_REL or SHT_RELA, and uses the dynamic symbol table, is
> - considered to be a dynamic reloc section. */
> +/* Return the size required for the dynamic reloc entries. Any loadable
> + section that was actually installed in the BFD, and has type SHT_REL
> + or SHT_RELA, and uses the dynamic symbol table, is considered to be a
> + dynamic reloc section. */
>
> long
> _bfd_elf_get_dynamic_reloc_upper_bound (bfd *abfd)
> @@ -5996,7 +5996,8 @@ _bfd_elf_get_dynamic_reloc_upper_bound (
>
> ret = sizeof (arelent *);
> for (s = abfd->sections; s != NULL; s = s->next)
> - if (elf_section_data (s)->this_hdr.sh_link == elf_dynsymtab (abfd)
> + if ((s->flags & SEC_LOAD) != 0
> + && elf_section_data (s)->this_hdr.sh_link == elf_dynsymtab (abfd)
> && (elf_section_data (s)->this_hdr.sh_type == SHT_REL
> || elf_section_data (s)->this_hdr.sh_type == SHT_RELA))
> ret += ((s->size / elf_section_data (s)->this_hdr.sh_entsize)
> @@ -6005,14 +6006,13 @@ _bfd_elf_get_dynamic_reloc_upper_bound (
> return ret;
> }
>
> -/* Canonicalize the dynamic relocation entries. Note that we return
> - the dynamic relocations as a single block, although they are
> - actually associated with particular sections; the interface, which
> - was designed for SunOS style shared libraries, expects that there
> - is only one set of dynamic relocs. Any section that was actually
> - installed in the BFD, and has type SHT_REL or SHT_RELA, and uses
> - the dynamic symbol table, is considered to be a dynamic reloc
> - section. */
> +/* Canonicalize the dynamic relocation entries. Note that we return the
> + dynamic relocations as a single block, although they are actually
> + associated with particular sections; the interface, which was
> + designed for SunOS style shared libraries, expects that there is only
> + one set of dynamic relocs. Any loadable section that was actually
> + installed in the BFD, and has type SHT_REL or SHT_RELA, and uses the
> + dynamic symbol table, is considered to be a dynamic reloc section. */
>
> long
> _bfd_elf_canonicalize_dynamic_reloc (bfd *abfd,
> @@ -6033,7 +6033,8 @@ _bfd_elf_canonicalize_dynamic_reloc (bfd
> ret = 0;
> for (s = abfd->sections; s != NULL; s = s->next)
> {
> - if (elf_section_data (s)->this_hdr.sh_link == elf_dynsymtab (abfd)
> + if ((s->flags & SEC_LOAD) != 0
> + && elf_section_data (s)->this_hdr.sh_link == elf_dynsymtab (abfd)
> && (elf_section_data (s)->this_hdr.sh_type == SHT_REL
> || elf_section_data (s)->this_hdr.sh_type == SHT_RELA))
> {
> @@ -7908,7 +7909,7 @@ long
> _bfd_elf_get_synthetic_symtab (bfd *abfd,
> long symcount ATTRIBUTE_UNUSED,
> asymbol **syms ATTRIBUTE_UNUSED,
> - long dynsymcount ATTRIBUTE_UNUSED,
> + long dynsymcount,
> asymbol **dynsyms,
> asymbol **ret)
> {
> @@ -7924,10 +7925,14 @@ _bfd_elf_get_synthetic_symtab (bfd *abfd
> char *names;
> asection *plt;
>
> + *ret = NULL;
> +
> if ((abfd->flags & (DYNAMIC | EXEC_P)) == 0)
> return 0;
>
> - *ret = NULL;
> + if (dynsymcount <= 0)
> + return 0;
> +
> if (!bed->plt_sym_val)
> return 0;
>
> Index: binutils/nm.c
> ===================================================================
> RCS file: /cvs/src/src/binutils/nm.c,v
> retrieving revision 1.43
> diff -u -p -r1.43 nm.c
> --- binutils/nm.c 14 Oct 2004 09:36:54 -0000 1.43
> +++ binutils/nm.c 25 Jan 2005 10:07:56 -0000
> @@ -1026,8 +1026,18 @@ display_rel_file (bfd *abfd, bfd *archiv
> }
> else
> {
> + long storage = bfd_get_dynamic_symtab_upper_bound (abfd);
> +
> static_count = symcount;
> static_syms = minisyms;
> +
> + if (storage > 0)
> + {
> + dyn_syms = xmalloc (storage);
> + dyn_count = bfd_canonicalize_dynamic_symtab (abfd, dyn_syms);
> + if (dyn_count < 0)
> + bfd_fatal (bfd_get_filename (abfd));
> + }
> }
> synth_count = bfd_get_synthetic_symtab (abfd, static_count, static_syms,
> dyn_count, dyn_syms, &synthsyms);
>
>
prev parent reply other threads:[~2005-01-26 20:33 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-01-24 21:51 Andrew Cagney
2005-01-24 21:53 ` Ian Lance Taylor
2005-01-24 22:24 ` Andrew Cagney
2005-01-25 9:54 ` Alan Modra
2005-01-25 10:30 ` Alan Modra
2005-01-26 20:33 ` Andrew Cagney [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=41F7FE5F.3060703@gnu.org \
--to=cagney@gnu.org \
--cc=amodra@bigpond.net.au \
--cc=binutils@sources.redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).