Re: [libgomp] No references to env.c -> no libgomp construction

[Alexander Monakov <amonakov at ispras dot ru>]

[redirecting from gcc@ to gcc-patches@ for patch submission]

On Wed, 30 Nov 2016, Sebastian Huber wrote:
> > On Tue, 29 Nov 2016, Sebastian Huber wrote:
> > >         * env.c: Split out ICV definitions into...
> > >         * icv.c: ...here (new file) and...
> > >         * icv-device.c: ...here. New file.
> > > 
> > > the env.c contains now only local symbols (at least for target *-rtems*-*):
> > > 
> > [...]
> > > 
> > > Thus the libgomp constructor is not linked in into executables.
> > 
> > Thanks for the report.  This issue affects only static libgomp.a (and not on
> > NVPTX where env.c is deliberately empty).
> > 
> > I think the minimal solution here is to #include <env.c> from icv.c instead of
> > compiling it separately (using <> inclusion rather than "" so in case of NVPTX
> > we pick up the empty config/nvptx/env.c from toplevel icv.c).
> > 
> > A slightly more involved but perhaps a preferable approach is to remove
> > config/nvptx/env.c, introduce LIBGOMP_OFFLOADED_ONLY macro, and use it to
> > guard inclusion of env.c from icv.c (which then can use the #include "env.c"
> > form).
> 
> I guess its sufficient to move
> 
> pthread_attr_t gomp_thread_attr;
> 
> from team.c (NVPTX seems to provide its own team.c) to env.c.  This generates
> a reference from team.c to env.c and the constructor is pulled in.

Well, yes, generally definitions of variables must be in the same translation
unit as the constructor that would initialize them -- so at the moment it's
wrong for both the ICV definitions and gomp_thread_attr to be defined elsewhere.

In reply to this message I'm posting 3 patches that solve the issue by moving
ICV definitions back to env.c and using new LIBGOMP_OFFLOADED_ONLY macro to
avoid environment-related functionality on nvptx.  I think it would be good to
move gomp_thread_attr to env.c too, but that's not a part of this patchset.

Thanks.
Alexander