[build, libgomp] Support Sun symbol versioning in libgomp, rev. 2

[Jakub Jelinek]

On Mon, Jul 05, 2010 at 06:58:22PM +0200, Rainer Orth wrote:
> Jakub Jelinek <jakub@redhat.com> writes:
> 
> > On Thu, Jul 01, 2010 at 05:45:45PM +0200, Rainer Orth wrote:
> >> Bootstrapped without regressions on Solaris 8 to 11, both SPARC and x86
> >> with Sun as and GNU as, Sun ld and GNU ld.
> >
> > Please also test on Linux (e.g. x86_64, i686, powerpc*) to make sure it
> > doesn't regress there.
> >
> > readelf -Ws $1 | sed -n '/\.symtab/,$d;/ UND /d;/@GLIBC_PRIVATE/d;/\(GLOBAL\|WEAK\|UNIQUE\)/p' | awk '{ if ($4 == "OBJECT") { printf "%s %s %s %s %s\n", $8, $4, $5, $6, $3 } else { printf "%s %s %s %s\n", $8, $4, $5, $6 }}' | LC_ALL=C sort -u
> >
> > should be identical before/after your patch.
> 
> I've discovered we have a single Linux machine running RHEL4, so I've
> bootstrapped without regressions on x86_64-unknown-linux-gnu.  The set
> of exported symbols is identical between a vanilla tree and my patch.

Thanks.

> Since libgomp uses the same target_os-based check for symbol versioning
> that has been inherited/copied from libstdc++-v3, there's no risk of
> running into the Darwin problem of silently accepting the ld -M option.
> 
> Ok now?

Ok.

	Jakub