This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: String table optimization?
- From: Jakub Jelinek <jakub at redhat dot com>
- To: Axel Kittenberger <Axel dot Kittenberger at maxxio dot com>
- Cc: Richard Henderson <rth at redhat dot com>, binutils at sources dot redhat dot com
- Date: Thu, 2 May 2002 09:32:12 +0200
- Subject: Re: String table optimization?
- References: <200204300748.JAA19467@merlin.gams.co.at> <20020430115117.A2965@redhat.com> <200205020653.IAA29403@merlin.gams.co.at>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Thu, May 02, 2002 at 08:53:58AM +0200, Axel Kittenberger wrote:
> On Tuesday 30 April 2002 20:51, Richard Henderson wrote:
> > On Tue, Apr 30, 2002 at 09:48:36AM +0200, Axel Kittenberger wrote:
> > > Do you know of a ld switch one can issue, or can someone explain why this
> > > systematicly could not be done safely?
> >
> > Update to current binutils and gcc 3.1 prelease, and this will happen.
>
> Is this only a binutils issue? I mean shouldn't it work with gcc 2.95.x also?
> I think gcc has only control over one object file, however linking them
> together should be a binutils specific issue, isn't it? Does it work with
> cross targets also?
The compiler must tell the linker what are strings (particularly strings
safe for this optimization, not all of them are, think about "foo\0bar").
> gcc 3.0 and upward is a little troublesome to use for linux kernels, glibc's
> and especially cross-compilers. I'm hesitating to upgrade over gcc 3 for this
> purpose, especially a prerelease.
Can you explain why is gcc 3.x troublesome to use for glibc?
Anyway, if you cannot use gcc 3.1, you can always backport the SHF_MERGE
changes to 2.95.x or use gcc-2.96-RH which supports this for a long time.
Jakub