https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82141
--- Comment #19 from Iain Sandoe <iains at gcc dot gnu.org> --- (In reply to Eric Botcazou from comment #18) > > As said in comment 3, I do (this PR) with darwin16: caused by r251781 > > ada/sem_ch7.adb, i.e., your change. > > This doesn't make much sense though since it only changes the set of symbols > declared public in packages. But it would be interesting to find out what > it really changes for the Rtsfind package. > > Since I have a working compiler on Darwin with this change: > > Target: x86_64-apple-darwin14.5.0 > Configured with: ../src/configure --enable-languages=ada,c,c++,objc > --enable-lto --with-stage1-ldflags=-static-libstdc++ > --with-boot-ldflags=-static-libstdc++ --disable-libcilkrts > --with-bugurl=URL:mailto:rep...@adacore.com --disable-nls > --without-libiconv-prefix --disable-libstdcxx-pch --disable-libada > --enable-checking=release --disable-multilib > --with-mpfr=/gnatmail/sandbox/wave/x86_64-darwin/mpfr_stable/install > --with-gmp=/gnatmail/sandbox/wave/x86_64-darwin/gmp_stable/install > --with-mpc=/gnatmail/sandbox/wave/x86_64-darwin/mpc_stable/install > --with-build-time-tools=/gnatmail/sandbox/wave/x86_64-darwin/gcc/build/ > buildtools/bin --prefix=/gnatmail/sandbox/wave/x86_64-darwin/gcc/pkg > --build=x86_64-apple-darwin14.5.0 > Thread model: posix > > I'm going to conduct some investigation on my side. * I suspect these are the key difference - since the fail is occurring in building libada, AFAICS - plus we don't typically disable multilibs for a general Darwin toolchain. In this case it seems to be the default multilib that's failing (so probably not the significant effect) --disable-libada --disable-multilib I'll try to repeat your case later - no spare cycles right now