http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59723
--- Comment #8 from rguenther at suse dot de <rguenther at suse dot de> --- On Thu, 9 Jan 2014, dominiq at lps dot ens.fr wrote: > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59723 > > --- Comment #7 from Dominique d'Humieres <dominiq at lps dot ens.fr> --- > > ... which implies that the GCC testsuite needs to be improved with regards > > to LTO testing as there was no new failure with the patch. Maybe ... but if you modify LTO code for a feature I'd expect you try LTO on a simple example using that feature ;) > Not GCC, but gfortran. I stumbled on this pr with > > make -k check-gfortran > RUNTESTFLAGS="--target_board=unix'{-m32/-flto,-m64/-flto}'" > > One may consider to add something '-Ofast -flto' to the set of options in > gfortran.dg. One always has to weight in testing time of course.