This is the mail archive of the
glibc-bugs@sourceware.org
mailing list for the glibc project.
[Bug math/19058] [x86_64] Link fail with -fopenmp and -flto
- From: "hjl.tools at gmail dot com" <sourceware-bugzilla at sourceware dot org>
- To: glibc-bugs at sourceware dot org
- Date: Wed, 10 Feb 2016 16:50:32 +0000
- Subject: [Bug math/19058] [x86_64] Link fail with -fopenmp and -flto
- Auto-submitted: auto-generated
- References: <bug-19058-131 at http dot sourceware dot org/bugzilla/>
https://sourceware.org/bugzilla/show_bug.cgi?id=19058
--- Comment #14 from H.J. Lu <hjl.tools at gmail dot com> ---
(In reply to joseph@codesourcery.com from comment #13)
> On Wed, 10 Feb 2016, hjl.tools at gmail dot com wrote:
>
> > https://sourceware.org/bugzilla/show_bug.cgi?id=19058
> >
> > --- Comment #12 from H.J. Lu <hjl.tools at gmail dot com> ---
> > (In reply to joseph@codesourcery.com from comment #11)
> > > The alias is working around a compiler limitation. The right thing to do
> > > long term is to get consensus on a compiler feature for assigning names to
> > > vector versions of functions independent of the set of scalar versions,
> > > not to make the ABI larger than necessary to work around the lack of such
> > > a feature.
> >
> > Should the workaround be in GCC then? Otherwise, it will be used
> > even when compiler works fine.
>
> Given a suitable new compiler feature in use, there will be no undefined
> references to the *_finite vector function names, and so none of the
> objects from libmvec_nonshared.a will be linked in.
This only fails with LTO. Why is libmvec_nonshared.a used in PR 19590 without
LTO?
--
You are receiving this mail because:
You are on the CC list for the bug.