This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [LTO] Committed merge feedback to the branch
- From: Richard Guenther <rguenther at suse dot de>
- To: Diego Novillo <dnovillo at google dot com>
- Cc: gcc-patches at gcc dot gnu dot org, Jason Merrill <jason at redhat dot com>, Mark Mitchell <mark at codesourcery dot com>, "Joseph S. Myers" <joseph at codesourcery dot com>
- Date: Fri, 2 Oct 2009 17:18:49 +0200 (CEST)
- Subject: Re: [LTO] Committed merge feedback to the branch
- References: <20091001193205.GA13840@google.com> <alpine.LNX.2.00.0910021158260.4520@zhemvz.fhfr.qr>
On Fri, 2 Oct 2009, Richard Guenther wrote:
> On Thu, 1 Oct 2009, Diego Novillo wrote:
>
> > This patch incorporates most of the feeback I've received from
> > the merge.
> >
> > In particular, this removes the need to make any further changes
> > in the C++ front end. We discussed an alternative way of dealing
> > with it directly in the middle end.
> >
> > When free_lang_data is walking down types and symbols, it ignores
> > any language-specific tree node. Since these are removed, it
> > does not make sense adding them to the work lists. This prevents
> > us from trying to mangle template-related symbols.
> >
> > The bulk of the changes are documentation, formatting and
> > diagnostic messages.
> >
> > The next change will only be the user-level documentation. After
> > that, we should be ready to merge.
> >
> > Richi, could you do one more round of SPEC testing? I'm hoping
> > that without the C++ hacks, we can get a clean run.
>
> loads of testsuite segfaults on i586 and
>
> lto1: fatal error: elf_update() failed: no error.^M
>
> on ppc and s390.
Those were caused by me enabling LFS for GCC but not for libelf,
doesn't happen with a clean tree which is fine.
Richard.