public inbox for binutils@sourceware.org
 help / color / mirror / Atom feed
From: Alan Modra <amodra@bigpond.net.au>
To: Etienne Lorrain <etienne_lorrain@yahoo.fr>
Cc: binutils@sourceware.org
Subject: Re: `.sym' referenced in section `reloc_sym' of file.o: defined in discarded section `.text.sym' of file.o
Date: Wed, 17 May 2006 11:34:00 -0000	[thread overview]
Message-ID: <20060517003519.GJ19700@bubble.grove.modra.org> (raw)
In-Reply-To: <20060516152841.79869.qmail@web26907.mail.ukl.yahoo.com>

On Tue, May 16, 2006 at 05:28:41PM +0200, Etienne Lorrain wrote:
> --- Alan Modra <amodra@bigpond.net.au> wrote:
> > On Tue, May 16, 2006 at 01:34:25PM +0200, Etienne Lorrain wrote:
> > >  My problem is when the function (for instance) linux_set_params is not used
> > > at all in the link process, and is discarded because I am using GCC
> > > -ffunction-sections and LD --gc-sections, it is still referenced in
> > > ".section reloc_paramcode_section"
> > 
> > What is special about reloc_paramcode_section?  ie. How are you managing
> > to confuse the linker into thinking the reference in that section is not
> > a normal use, which should result in linux_set_params being kept?
> > 
> > Do you have a small self-contained testcase?
> 
>   I do not know what is special about reloc_paramcode_section, but I noticed

It's special because it is non-alloc, non-load.  --gc-sections does not
remove such sections, but does not treat their relocs specially.
ie. unless reloc_paramcode_section is itself referenced from somewhere,
its relocs will not be examined to see whether other sections referenced
by it should be kept.

See http://sources.redhat.com/ml/binutils/2004-08/msg00178.html  Perhaps
I should have restricted that change to sections without relocs.  That
would still keep sections like .comment and .note.GNU-stack, but drop
your "special" section.  I'm applying the following:

	* elflink.c (elf_gc_sweep): Don't specially keep non-alloc,
	non-load sections if they have relocs.

Index: bfd/elflink.c
===================================================================
RCS file: /cvs/src/src/bfd/elflink.c,v
retrieving revision 1.213
diff -u -p -r1.213 elflink.c
--- bfd/elflink.c	11 May 2006 15:55:40 -0000	1.213
+++ bfd/elflink.c	17 May 2006 00:28:23 -0000
@@ -8965,7 +8965,7 @@ elf_gc_sweep (bfd *abfd, struct bfd_link
 	{
 	  /* Keep debug and special sections.  */
 	  if ((o->flags & (SEC_DEBUGGING | SEC_LINKER_CREATED)) != 0
-	      || (o->flags & (SEC_ALLOC | SEC_LOAD)) == 0)
+	      || (o->flags & (SEC_ALLOC | SEC_LOAD | SEC_RELOC)) == 0)
 	    o->gc_mark = 1;
 
 	  if (o->gc_mark)


>  the message do not appear if --no-check-sections is not a parameter of ld !!!

That would be because you are hitting a fatal error without
--no-check-sections.

> ld: error: no memory region specified for loadable section `.rel.dyn'

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

  reply	other threads:[~2006-05-17  0:35 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-05-16 15:28 Etienne Lorrain
2006-05-16 22:39 ` Alan Modra
2006-05-16 22:47   ` Etienne Lorrain
2006-05-17 11:34     ` Alan Modra [this message]
2006-05-17 15:32       ` Fix regression introduced 2006-04-21 Alan Modra
2006-05-17 16:19       ` `.sym' referenced in section `reloc_sym' of file.o: defined in discarded section `.text.sym' of file.o Etienne Lorrain
2006-05-19 11:50         ` Alan Modra
2006-05-19 15:30           ` Etienne Lorrain
2006-05-19 15:37           ` Alan Modra
2006-05-19 15:57           ` Etienne Lorrain
2006-05-20 13:56             ` Alan Modra
2006-05-16 22:37 Etienne Lorrain

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=20060517003519.GJ19700@bubble.grove.modra.org \
    --to=amodra@bigpond.net.au \
    --cc=binutils@sourceware.org \
    --cc=etienne_lorrain@yahoo.fr \
    /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).