https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82686
--- Comment #6 from Dennis Clarke <dclarke at blastwave dot org> ---
Actually first thing I did was remove a few options from configure stage
such that I could at least get past this small bump in the road :
--enable-multiarch and --enable-multilib removed.
Configure went fine and the bootstrap went past the previous gc bump and then
crashed into some sort of an ada problem, again, in stage 3 :
$ /usr/bin/time -p /usr/bin/nice -n +19 /usr/local/bin/gmake -k 4
.
.
. a long while later
.
.
/usr/local/bin/ranlib rts_32/libgnarl.a
/usr/local/bin/ar rc rts_32/libgmem.a \
rts_32/memtrack.o
/usr/local/bin/ranlib rts_32/libgmem.a
chmod a-wx rts_32/*.ali
touch ../stamp-gnatlib-rts_32
gmake[7]: Leaving directory
'/usr/local/build/gcc-7.2.0_linux_4.13.0-1-powerpc64.004/gcc/ada'
mv -f libgna*.so rts_32
mv -f libgnat_pic.a rts_32
mv -f libgnarl_pic.a rts_32
gmake[6]: Leaving directory
'/usr/local/build/gcc-7.2.0_linux_4.13.0-1-powerpc64.004/gcc/ada'
gmake[5]: Leaving directory
'/usr/local/build/gcc-7.2.0_linux_4.13.0-1-powerpc64.004/gcc/ada'
rm -rf adainclude
rm -rf adalib
ln -s ../../.././gcc/ada/rts_32 adainclude
ln -s ../../.././gcc/ada/rts_32 adalib
true DO=all multi-do # /usr/local/bin/gmake
gmake[4]: Leaving directory
'/usr/local/build/gcc-7.2.0_linux_4.13.0-1-powerpc64.004/powerpc64-unknown-linux-gnu/32/libada'
gmake[3]: Leaving directory
'/usr/local/build/gcc-7.2.0_linux_4.13.0-1-powerpc64.004/powerpc64-unknown-linux-gnu/libada'
gmake[2]: Leaving directory
'/usr/local/build/gcc-7.2.0_linux_4.13.0-1-powerpc64.004/powerpc64-unknown-linux-gnu/libada'
gmake[1]: Leaving directory
'/usr/local/build/gcc-7.2.0_linux_4.13.0-1-powerpc64.004'
gmake: *** [Makefile:941: all] Error 2
Command exited with non-zero status 2
real 12929.78
user 41765.84
sys 2214.94
nix_$
nix_$
nix_$ cat stage_current
stage3
nix_$
So there really are problems in this bootstrap process and it isn't just
a new requirement for gc. In any case I will remove "ada" from the
language list and try again without multilib or multiarch and see what
happens. I expect that the need for multiple arch builds of gc can be
sorted out after I see at least a single arch complete, once.
