This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch] compile libgomp.c++/pr30703.C with -shared-libgcc


Mike Stump <mrs@apple.com> writes:

> Ian, I was wondering what your take on this as a shared libgcc
> maintainer?  I think it borders on obvious, but there is some concern
> that -shared-libgcc might not somehow work on some platforms.

I believe that -shared-libgcc will fail if gcc is configured with
--disable-shared.  -shared-libgcc will happily pass -lgcc_s, but there
won't be any such library.

Since this is just for testing purposes how about just doing a static
link?

Ian

> On Mar 15, 2007, at 12:47 PM, Mike Stump wrote:
> 
> > On Mar 15, 2007, at 6:50 AM, Jack Howarth wrote:
> >>    The attached patch forces the libgomp.c++/pr30703 testsuite test
> >> to compile with the shared libgcc to prevent false failures on
> >> platforms
> >> like Darwin. This is required because the dejagnu harness used for
> >> the libgomp.c++ tests doesn't compile those tests with the g++
> >> compiler. Okay for trunk and 4.2 branch?
> >
> > I reviewed the patch and the gcc documentation and it looks correct
> > enough to add that flag.  I think a better way to fix this would be
> > to instead do:
> >
> > Doing diffs in libgomp:
> > --- libgomp/testsuite/libgomp.c++/c++.exp.~1~	2006-11-16
> > 23:18:00.000000000 -0800
> > +++ libgomp/testsuite/libgomp.c++/c++.exp	2007-03-15
> > 12:42:02.000000000 -0700
> > @@ -1,6 +1,6 @@
> >  set lang_library_path	"../libstdc++-v3/src/.libs"
> >  set lang_test_file	"${lang_library_path}/libstdc++.a"
> > -set lang_link_flags	"-lstdc++"
> > +set lang_link_flags	"-shared-libgcc -lstdc++"
> >
> >  load_lib libgomp-dg.exp
> >
> > --------------
> >
> > The idea is to put the fix next to the thing that is broken, that
> > way it is impossible for another case to break.  Someone want to
> > agree with my patch and reasoning?  If so, I'll Ok that version.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]