This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: RFC libgomp/testsuite/libgomp.fortran
In article <20090902234241.GA45546@troutmask.apl.washington.edu>,
Steve Kargl<sgk@troutmask.apl.washington.edu> writes:
>> > As a gfortran maintainer and FreeBSD user, I've always considered these
>> > failures to be bogus. The inclusion of -lgomp and -pthread are tied
>> > to -fopenmp. If you look at the failures, you'll see an ordering of
>> > '-fopenmp -O -fno-openmp', which correctly disables the inclusion
>> > of -lgomp and -pthread. The only problem is that
>> > troutmask:sgk[251] find . -name \*.exp | xargs grep lgomp
>> > ./lib/libgomp.exp: lappend ALWAYS_CFLAGS "ldflags=-lgomp"
>> > so dejagnu overrides -fno-openmp and adds -lgomp without adding
>> > -pthread.
>> > Your changes look correct to me, but I can't approve patches
>> > in the libgomp hierarchy.
>> What would happen if we removed 'lappend ALWAYS_CFLAGS "ldflags=-lgomp"'
>> from libgomp.exp? That seems like that would be the 'right' fix.
> I don't know. Given that Jakub, Diego, and RTH wrote gomp,
> I assume that they had a reason for this to be in libgomp.exp.
For sample size one (port == i386-*-freebsd7*), the proposed removal
of: lappend ALWAYS_CFLAGS "ldflags=-lgomp"
"works" as well as my proposed patch to port-selectively add -pthread
to the -fno-openmp cases. I also verfied that every line which had
-lgomp also had (and still has) -fopenmp (which implies -lgomp).
The only remaining concern are those four tests that I patched to add
a port-selective -pthread have -fopenmp followed by -fno-openmp. Do
those four cases need -lgomp to properly test some run-time aspects?
I don't think so, but I don't really know fortran anymore.
Regards,
Loren