This is the mail archive of the
mailing list for the GCC project.
Re: PATCH: Use COMDAT section group instead of gnu.linkonce
- From: "H. J. Lu" <hjl at lucon dot org>
- To: Zack Weinberg <zack at codesourcery dot com>
- Cc: Mike Stump <mrs at apple dot com>, gcc-patches at gcc dot gnu dot org,Jim Wilson <wilson at specifixinc dot com>, gcc at gcc dot gnu dot org
- Date: Wed, 5 May 2004 11:53:02 -0700
- Subject: Re: PATCH: Use COMDAT section group instead of gnu.linkonce
- References: <20040504054619.GA5965@lucon.org> <4075E063-9DFC-11D8-B137-003065A77310@apple.com> <20040505182649.GA8969@lucon.org> <email@example.com>
On Wed, May 05, 2004 at 11:39:25AM -0700, Zack Weinberg wrote:
> "H. J. Lu" <firstname.lastname@example.org> writes:
> > On Tue, May 04, 2004 at 11:52:54AM -0700, Mike Stump wrote:
> >> On Monday, May 3, 2004, at 10:46 PM, H. J. Lu wrote:
> >> >Should we start moving toward COMDAT section group? It will help fix
> >> >DWARF2 problems.
> >> We are in favor of this. Our platform doesn't yet have group COMDAT,
> >> but, that is the way we know we need to go. Thanks.
> > This patch enables COMDAT group. It even works with
> > const int foo __attribute__((section(".gnu.linkonce.r.foo")));
> > There are no regressions on ia32, ia64 and x86-64. You just need to
> > apply
> > http://sources.redhat.com/ml/binutils/2004-05/msg00022.html
> > on top of the current binutils in CVS.
> Uh, surely there is a better way to do this than a long chain of
> strcmp() operations on the linkonce section name?
I want to keep my change to minimal and support existing source code.
The majority of .gnu.linkonce.* name will match one of them. I can
change it a little bit. But I don't think I can totally avoid it. Any